Before, the TThreadedObject had to guess 64 slots were enough for the use case of TTreeProcessorMT - and with 256 cores that was not the case.
Be explicit about the use of IMT for TThreadedObject by providing the new argument.
IMT is enabled before the construction of the TThreadedObject (and thus TThreadedObject now knows how many slots to expect), because the pool is constructed before.
If the order ever gets changes, the TThreadedObject will complain as it is now told to expect IMT usage.

A future TTreeProcessorMT ctor overload might provide a specific thread count, and can pass it down to TThreadedObject.

It is ambiguous whether TThreadedObject should rely on IMT, or whether it is used by user scheduling.
The default of 64 threads was good enough for user scheduling - unless the user needs more threads.
The default of 64 threads is also good enough for IMT, when IMT was not yet enabled - unless IMT ends up using more slots.
Thus, the silent default value is often misleading, and setting it through a static before construction is not exactly cplusplusy.

Instead, plovide a constructor tag that passed the number of slots - or "IMT" if that is what is wanted.
Add a warning if IMT is assumed but IMT is off, rather than failing, if IMT later needs more than fgMaxSlots slots.

In my (very artificial) benchmark, I see a 15% reduction in jitting
time thanks to this change.

* Remove ROOT.ROOT calls in the Python tutorials except for
  EnableImplicitMT to keep backwards compatibility until 6.24
* Use rdfentry_ instead of tdfentry_
* Use df for variables instead of tdf

When a NormRange with overlapping ranges was given, the plotting range
would be ignored.

[ROOT-10551] Plotting with disjoint ranges can be tricky, since the
normalisation of the PDF depends on its plotting range. The new tutorial
rf212 shows how to use NormRange to obtain identical normalisations.

When users set ranges that are not known by any of the fit observables,
there will now be a warning.

[ROOT-10550] When a ranged fit is repeated, RooAbsOptTestStatistic would
attach the fit range to the PDF for each iteration. This makes the fit
range string grow with every iteration.

When not setting CreateConfBelt() for FeldmanCousins or
NeymanConstruction, GetConfidenceBelt() returns nullptr instead of a
broken belt. The documentation also reflects this now.

[ROOT-4060] HistFactory uses the RooFactoryWSTool to create the index category of a
Simultaneous PDF. For this to work, the first channel name cannot start
with a digit, and channel names cannot contain ','.

[ROOT-7578] Is a duplicate of this issue.

[ROOT-8033] When not choosing anything, constraints for statistical
errors of Monte Carlo simulations defaulted to Gaussian. Now, they will
default to Poisson, as promised in the HistFactory documentation.
https://cdsweb.cern.ch/record/1456844/files/CERN-OPEN-2012-016.pdf

- [ROOT-5311] HistFactory is spamming messages to cout when hist2workspace runs.
- A new message topic HistFactory was added to RooFit's message streams,
  which allows to collect and also suppress these messages using e.g.
     RooMsgService::instance().getStream(2).minLevel = RooFit::PROGRESS;
- Add HFMsgService header with shortcut definitions that route
  HistFactory messages through RooMsgService.
- ObjectHandling messages are now automatically suppressed when HistFactory runs.
- [ROOT-9365] Documentation of HistFactory classes and ParamHistFunc improved a bit.
- Make hist2workspace much less verbose.
  - Reduce default message level of hist2workspace.
  - Add -v and -vv options for hist2workspace for increasing verbosity.

- LocalChangeMsgLevel so far could only add message topics to stream 0.
Now, one can locally add and remove message topics, and also change the
message level on all streams.
- Fix typos in RooMsgService shortcuts.

Improve TH2::FitSlicesX/Y and TH3::FitSlicesZ by adding support for axis range when they are defined.
A similar behaviour as in case of projection is used.

Improve also the code of TH3::FitSlicesZ to use internally TH3::ProjectionZ

This fixes the JIRA issue ROOT-10146

A user now can define an axis range in Y for FitSlicesX and an axis range in X for FitSlicesY.
The histograms contained the parameters of the fitted slices will be then defined using that range.

This fixes the TH2 part of ROOT-10146

Fix the case in TH2::FitSlicesX/Y when the projected histograms have a very small number of effective entries and int(nentries) = 0 .

Use option "N" do not plot, do not store fit function when fitting the internal histograms in TH2::FitSlicesX/Y

This fixes ROOT-10601

Removing TClingRuntime.h, string.h, stddef.h, and stdlib.h.

Add CrossValidationFoldResult in the LinkDef. This fixes the Multiprocess running of CrossValidation

* Remove empty lines
* Remove unnecessary braces in expressions
* Add missing | for df105

The current logic of pseudo-automatic modulemap discovery is a little fragile
as it mounts a virtual file named module.modulemap in the location where the
a give system library header is found (cuda.h, etc). However, if the libraries
are squashed into a single folder then we end up mounting a file over another
virtual file. This should be a hard error, however, on systems such as OSX some
libraries already come with modulemap files and we should just not mount our
predefined ones. This makes it very difficult to distinguish when to issue the
hard error.

While this patch is not a solution for the general when the modulemaps are
automatically discovered (-fimplicit-module-maps), we can use the real file
stems in cases where we are called with -fno-implicit-module-maps.

As a side effect this fixes the *non-recommended* squashing of library header
files into a single folder.

After investigating the bug reported here

https://sft.its.cern.ch/jira/browse/ROOT-10626

we noticed that the deprecated variables
PYTHON_EXECUTABLE
PYTHON_INCLUDE_DIRS
PYTHON_LIBRARIES
PYTHON_VERSION_MAJOR
PYTHON_VERSION_MINOR
PYTHON_VERSION_STRING
NUMPY_FOUND
NUMPY_INCLUDE_DIRS
are not set correctly in the CMakeCache. This should fix the problem.

Raise a warning if requesting a specific number of threads but the pool
has already been initialized with a different number.