These files are unused and can be removed.

This should fix this warning in the ROOT nightlies on the mac13 node:

```
core/clib/src/mmtrace.c:32:14: warning: a function declaration without a prototype is deprecated in all versions of C and is not supported in C2x [-Wdeprecated-non-prototype]
extern char *getenv ();
```

https://lcgapp-services.cern.ch/root-jenkins/view/ROOT%20Nightly/job/root-nightly-master/LABEL=mac13,SPEC=cxx20,V=master/lastBuild/parsed_console/

This should fix some warnings in the CI:

https://lcgapp-services.cern.ch/root-jenkins/view/ROOT%20Nightly/job/root-nightly-master/LABEL=mac11arm,SPEC=noimt,V=master/lastBuild/parsed_console/

Also, the equality operator is replaced with the free function version
because this is preferred in C++20.

This is a follow-up to 82d17fb6, fixing these warnings seen in the
nightlies (Ubuntu 20.04):

https://lcgapp-services.cern.ch/root-jenkins/view/ROOT%20Nightly/job/root-nightly-master/LABEL=ROOT-ubuntu2004-clang,SPEC=soversion,V=master/lastBuild/parsed_console/

```
/home/sftnight/build/night/LABEL/ROOT-ubuntu2004-clang/SPEC/soversion/V/master/build/roofit/roofitcore/G__RooFitCore.cxx:8047:11: warning: misleading indentation; statement is not part of the previous 'if' [-Wmisleading-indentation]
          if(onfile._vecEL) _vecEL = *onfile._vecEL; \
          ^
/home/sftnight/build/night/LABEL/ROOT-ubuntu2004-clang/SPEC/soversion/V/master/build/roofit/roofitcore/G__RooFitCore.cxx:8046:7: note: previous statement is here
      if(onfile._vecE) _vecE = *onfile._vecE;    \
      ^

/home/sftnight/build/night/LABEL/ROOT-ubuntu2004-clang/SPEC/soversion/V/master/build/roofit/roofitcore/G__RooFitCore.cxx:8048:11: warning: misleading indentation; statement is not part of the previous 'if' [-Wmisleading-indentation]
          if(onfile._vecEH) _vecEH = *onfile._vecEH;
          ^
/home/sftnight/build/night/LABEL/ROOT-ubuntu2004-clang/SPEC/soversion/V/master/build/roofit/roofitcore/G__RooFitCore.cxx:8047:11: note: previous statement is here
          if(onfile._vecEL) _vecEL = *onfile._vecEL; \
          ^
```

isysroot influences where clang will pick up libc++. Without this, and with
Xcode 14.3, cling will use libc++ from Xcode (or the command line tools) rather
than stdc++ from the macOS SDK, as clang would normally use. Passing the isysroot
(which point to the SDK) fixes this.

This solves build errors such as:
```
While building module 'Core':
While building module 'std' imported from input_line_1:1:
In file included from <module-includes>:17:
In file included from /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/../include/c++/v1/math.h:309:
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/../include/c++/v1/type_traits:462:1: error: cannot template a using declaration
using _BoolConstant _LIBCPP_NODEBUG_TYPE = integral_constant<bool, _Val>;
^
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/../include/c++/v1/type_traits:462:21: error: C++ requires a type specifier for all declarations
using _BoolConstant _LIBCPP_NODEBUG_TYPE = integral_constant<bool, _Val>;
                    ^
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/../include/c++/v1/type_traits:462:68: error: use of undeclared identifier '_Val'
using _BoolConstant _LIBCPP_NODEBUG_TYPE = integral_constant<bool, _Val>;
                                                                   ^
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/../include/c++/v1/type_traits:462:73: error: expected '(' for function-style cast or type construction
using _BoolConstant _LIBCPP_NODEBUG_TYPE = integral_constant<bool, _Val>;
                                           ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
```

The C++ Standard requires a complete type T when using any members of
`vector<T>`, see
https://eel.is/c++draft/vector#overview-4.

This only breaks with latest libc++ in C++20 mode and does not show up
in common configurations.
We have an internal experimental configuration that discovered this.

Reviewed By: alexfh

Differential Revision: https://reviews.llvm.org/D131595

Some methods of json::Array require json::Value to be completely defined, so
they can't be defined in-class. Fix that by defining them out of class.

Fix #55780

In 13670912, the class version of the `RooAbsReal` class was
incremented.

As explained in #8791, there are sometimes warnings in the IO of derived
classes if their version number is not increased as well.

Increasing the class version of RooRealVar indeed fixes this warning
that one gets right now when reading old workspaces:

```
Warning in <TStreamerInfo::CompareContent>: The following data member of
the on-file layout version 9 of class 'RooRealVar' differs from
the in-memory layout version 9:
   RooAbsBinning _binning; //
vs
   unique_ptr<RooAbsBinning,default_delete<RooAbsBinning> > _binning;
```

Thanks to @will-cern for catching this problem!

Like this, the code that uses the JSON interface in RooFit HS3 can be
made less verbose.

This fixes the remaining differences in the fit result from old and new
HistFactory.

This commit works around a problem in RooStats/HistFactory that was only
fixed in ROOT 6.30 (commit `ee0b06a3`): until then, the weight
variable of the observerd dataset, called "weightVar", was added to the
observables. Therefore, it also got added to the Asimov dataset. But the
Asimov has its own weight variable, called "binWeightAsimov", making
"weightVar" an actual observable in the Asimov data. But this is only by
accident and should be removed.

Check if any exported RooDataSet actually represents a binned dataset,
and then import it like a RooDataHist. This happens frequently when
people create combined RooDataSets from binned data to fit HistFactory
models. In this case, it doesn't make sense to export them like an
unbinned dataset, because the coordinates are redundant information with
the binning. We only do this for 1D data for now.

When having combined datasets that are meant for a simultaenous fit,
it's more efficient to use the associated RooSimultaneous to figure out
which observables need to be stripped off the component datasets.

This also fixes a problem with RooDataSets with single entries being
exported incorrectly.

Instead of guessing the number of paramters for the model represented by
RooFuncWrapper, we now introduce a parameter size function which can be
used to declare gradient output arrays of the right size.

Closes JIRA issue 6039:
https://sft.its.cern.ch/jira/browse/ROOT-6039

In the issue, it was suggested to make that the default, but I think
this is a too drastic change. People often have long variable titles,
and putting these by default in the plot labels might mess up the plots.

As a compromise, this commit suggests an easy option to change the
behavior of `paramOn()`, just using `T` in the format string instead of
`N`.

The `Format(const char*, int)` command arg for `RooAbsPdf::paramOn()` is
long superseeded by `Format(const char*, ...)`, which takes additional
command arguments instead of a single additional integer to specify the
number of significant digits.

This should obviously be `CUDA_TOOLKIT_ROOT` with an L.

I don't know if this typo has any negative consequences, but it's maybe
better to fix it before there are bugs creeping up because of that.

* [ci][windows] Enable mathmore on Windows

* [ci][windows] Remove duplicates wrt global.txt

* [c++20] Add spaceship operator <=> in TString

In the following code (from `TFormula.cxx`):
```
   map< std::pair<TString,Int_t>, pair<TString,TString> > functions;
   pair<TString, Int_t> key = make_pair(funName, dim);
   if (functions.find(key) == functions.end()) {
```
`functions.find(key)` always returns `functions.end()`. Adding the spaceship operator `<=>` in TString fixes the issue.
This fixes also many failing tests with `std:c++20`

* Use __cpp_impl_three_way_comparison instead of __cplusplus version (thanks Bernhard Manfred Gruber!)

* Apply suggestion from Bernhard Manfred Gruber

* Make sure we only include <compare> with at least std=c++20

* Implement the proper way (avoid hidden friend) thanks Bernhard

* Prevent calling 'operator <=>' recursively

Prevent the following warning on Windows:
```
C:\root-dev\git\master\core\base\inc\TString.h(175,16): warning C5232: in C++20 this comparison calls 'std::strong_ordering operator <=>(const TString &,const TString &)' recursively
```

The expected output of Display was set to what Display returned,
but what Display returned was wrong (because of #12597).

Before this commit, RDisplay forgot to reset the state of the
collection representation between events.

This fixes #12597.

The implementations of the virtual `RooAbsL::getParameters()` either
returned an owning of a non-owning pointer, depending on the
implementation class. RooSubsidiaryL returned a non-owning pointer, and
the default implementation an owning pointer.

The caller code of course doesn't know what to do with this, resulting
in memory leaks.

To fix this problem, the interface is suggested to return
`std::unique_ptr<RooArgSet>`. Changing this interface now is okay
because it has not been adopted by the users yet (it's from the new test
statistic classes introduced in ROOT 6.26).

As noted in commit 538396a8 some weeks ago, Clang 15 warns that the
variable v may be used uninitialized because it does not understand
the pointer hackery to read individual bytes.

These are just technical changes to avoid code duplication in the RooFit
JSON IO.

It is only necessary to set the `combined_data_name` of a simultaneous
pdf that gets exported to JSON when you also have a ModelConfig that
will get exported to an `analysis` object in the JSON.

Just like with the combined datasets, the JSON needs to contain all the
necessary information to reconstruct simultaneous pdfs.

Now, the unit test that validates the IO of a simultaneous model can
also be fully activated.

It was not good to first export RooHistPdf to TH1 before writing them to
the JSON. The problem is that the exported TH1 is normalized, and
rounding errors in the normalization change the information.

Now, the unit test for the RooHistPdf IO can also contain more random
numbers, which was not possible before because of the precision errors.

Technical changes to reduce code duplication.

To correctly import and export combined datasets from RooFit, the JSON
needs to contain all the information to rebuild the combined dataset
from the components.

Co-authored-by: Wim Lavrijsen <WLavrijsen@lbl.gov>

We only have release builds for incrementals, i.e. only release
builds use artifacts, so there is no point in building debug artifacts.