AthDerivation (CentOS 7) Fixes, 21.2 branch (2020.03.05.)

These are the things that I had to do to compile AthDerivation for the x86_64-centos7-gcc62-opt platform using Ninja as the build system. For the current x86_64-centos7-gcc62-opt nightly's issues you can have a look at, for instance:

http://atlas-computing.web.cern.ch/atlas-computing/links/distDirectory/gitwww/GITWebArea/nightlies/21.2/2020-03-05T0018/AthDerivation/x86_64-centos7-gcc62-opt/AthDerivation/

I had to do the following:

• To get the Ninja build started, I had to:
  • Stop xAODRootAccess from installing checkxAOD.py in anything but the AnalysisBase project. Every Athena based project provides this same script from PyUtils/bin/checkxAOD.py as well, and I decided to have that one installed in those instead.
  • I had to fix a typo in the TauHistUtils package's CMake configuration.
  • Remove DerivationFrameworkPhysicsValidation/test/test_mc16PHYSVAL.sh. A file with the name test_mc16PHYSVAL.sh is being installed from the following place as well:

https://gitlab.cern.ch/atlas/athena/blob/21.2/PhysicsAnalysis/DerivationFramework/DerivationFrameworkART/DerivationFrameworkPhysicsValidationART/test/test_mc16PHYSVAL.sh

@tcuhadar may correct me, but I guess that is the "more modern" place for this script, right?

• Fix a missing dependency on GSL in TruthRivetTools. On the nightly/CI nodes GSL must be picked up during the build from /usr. Very incorrectly... On my private node GSL is not installed on the system, so this configuration error was revealed.
• As discussed in ATLINFR-3380, it turns out that the x86_64-centos7-gcc62-opt build of tdaq-common-02-02-00 is built against UUID from /usr. So I had to force AthDerivation to not pick it up from LCG_88 either.
• Based on @tcuhadar's feedback I also moved the remaining ART script out of DerivationFrameworkPhysicsValidation.