* [skip] norm shadows a declaration in the 'std' namespace

* Fix format

Prevent the parallel builds of tests from roottest to trigger rebuilding of rootcling_stage1.exe on Windows

* Fix several warning GC8D28483 (inconsistent-missing-override)

* Remove the virtual keyword for the overriden methods

* Modified the logic of MapReduce and Map of the TThreadExecutor which were
initially broken for the TThreadExecutor;
* Added corresponding tests.

Fix https://github.com/root-project/root/issues/10404

In TUnixSystem.cxx's dynpath, the full result is know only when the TCling object is created but
it is called several before that.  Part of the work is to get information from the dynamic
linker via a call to popen("LD_DEBUG=libs LD_PRELOAD=DOESNOTEXIST ls 2>&1").

Enhance the code so this call (and also to call to getenv) are only done once.

`long double` is the most precise type for evaluating the TTreeFormula expressions.
For some platforms - notably Apple M1 - `long double` has the same precision as `double`,
and thus not enough bits to represent large numbers, as shown by roottest/tree/index/runindex64.C
failing on M1. The most reasonable way to deal with this is to warn if any of the relevant
indexing values exceeds the precision of `long double`.

For that, check whether an expression with a value 1 closer to zero
would result in the same result, i.e. this value is saturated.

* Fix DF examples to make them work as notebooks

* Add missing space

* python exec in filter.cxx

* use "authors"

* notebook does not make sense for these tutorials.

Co-authored-by: Axel Naumann <Axel.Naumann@cern.ch>
Co-authored-by: Philippe Canal <pcanal@fnal.gov>

Most commonly seen on ppc64le. Backtrace:

===========================================================
The lines below might hint at the cause of the crash.
You may get help by asking at the ROOT forum https://root.cern/forum
Only if you are really convinced it is a bug in ROOT then please submit a
report at https://root.cern/bugs Please post the ENTIRE stack trace
from above as an attachment in addition to anything else
that might help us fixing this issue.
===========================================================
 #11 ROOT::Experimental::RColor::toHex[abi:cxx11](unsigned char) (v=<optimized out>) at /usr/include/c++/11/ext/new_allocator.h:82
 #12 0x00007fff90c220ec in ROOT::Experimental::RColor::SetRGB (this=0x7fffeadf5d10, r=<optimized out>, g=<optimized out>, b=<optimized out>) at /usr/include/c++/11/ext/new_allocator.h:89

otherwise .typedef command shows weird format

The last chunk of partial results of MapReduce with std::vectors had the
size of all other chunks. This is dangerous, since the default vector
values might cause unexpected results of the Reduce function.

Test added to verify the correctness.

* make .Class the verbose equivalent of .class and take into account name passed to .Class. As it was the case in CINT.
* merge similar functions into one with a verbose arg

fix indentation

* fix indent as with cling
* adapt indent in shortcuts too

The `resize` called a few lines above initializes elements as
necessary. We don't need to call GenerateValue on those memory
locations (which performs another placement new on them), rather
a CaptureValue is enough.

* Clarify DoProject2D return pointer ownership

Fixes https://github.com/root-project/root/issues/10400

* Clarify only delete if AddDirectory is false

* apply lmoneta's suggestions

Co-authored-by: Javier Lopez-Gomez <javier.lopez.gomez@cern.ch>

* [skip-ci] df105 nbconvert fix

* Include Enric suggestion

Adds a sanity check in a TTree-based distributed analysis. In every
task, get the number of entries of each tree and the number of processed
entries in that task. Merge these values in the reduce phase. When the
final result is returned to the client, we have a struct holding the
number of entries in each tree that was open in the distributed workers,
and the total number of entries processed overall. Perform the following
checks:
- The identifiers filename?#treename are the same for the trees that
  were open in the workers and the trees that were passed in input by
  the user before starting the execution.
- The total number of processed entries is the same as the sum of all
  entries in all trees. We need to also take into account multiple times
  the same tree if the user requested it.

Adapt old tests to new logic, move all tests that check ranges in the
same file. Add new tests

This removes the need to open input files of the dataset in the local
machine. Now each distributed task will receive a list of percentages
that should be considered for each file. The files needed for a certain
task are opened directly in that task. The application constructs these
percentages locally by considering each file as an entity that can be
split in 100 parts. The starting and ending percentage of a certain file
is then translated to starting and ending clusters for the same file
proportionally.

Co-authored-by: Enrico Guiraud <enrico.guiraud@cern.ch>
Co-authored-by: Enric Tejedor Saavedra <enric.tejedor.saavedra@cern.ch>