Now that DerivationFramework::FloatCompressionAlg doesn't need a change
in this class, I restored it to its previous setup. Apart from making the
two constant variables of the class static. (Which they really should
have been from the start...)

Instead of teaching a single xAOD::FloatCompressor object how to compress
float variables to different precisions, I thought it would make more sense
for DerivationFramework::FloatCompressionAlg to just create as many
xAOD::FloatCompressor objects as it needs.

add 2017 and 2018 triggers to ART config files (ANALYSISTO-780)

See merge request atlas/athena!22358

SUSYTools: Update actualMu file for 2018 and workaround for ATLSUSYSW-463

See merge request atlas/athena!22349

HIGG1D2 - remove memory leak

See merge request atlas/athena!22309

AnaAlgorithm Metadata Access, 21.2 branch (2019.04.01.)

See merge request atlas/athena!22302

With Nils's updates in place, it only took a small further tweak to EL::TEventModule
to get the intended behaviour.

At the same time extended the unit test for this behaviour to make sure that algorithms
behave as we want them to.

Unfortunately I did not resolve the merge conflict correctly. :-(
I should have been more careful...

Had to resolve a conflict in EventLoopTest/CMakeLists.txt. The recently added
EventLoopGrid test interfered with the cleanup that I'm trying to do there.

fix to IsolationConditionCombined::accept for incoming StrObj from CloseBy...

See merge request atlas/athena!22339

delay algorithm::intialize until *after* the first file is open

See merge request atlas/athena!22317

fix to IsolationConditionCombined::accept for incoming StrObj from CloseBy tool - wasn't correctly implemented

add option to run the jet resolution for muons and electrons for the METSignificance

See merge request atlas/athena!22299

Bugfix in VxCandidateXAODVertex tool

See merge request atlas/athena!22310

This is to allow users to query meta-data inside initialize.  Right
now this also allows users to retrieve event-data inside initialize
(unless the first file contains no events), but that will be disabled
at a later point.

This now no longer allows to run on samples without any files in them.
I could potentially somehow work around that, but there isn't a strong
use-case for that and it would mean that user code also would need
some specialized handling for that case.

Also removed the Module::postInitialize method since the one module
using it could also run in onInitialize instead, thereby simplifying
the module interface somewhat.

Implementation of insitu JES calibration for Large Jets

See merge request atlas/athena!22065

AFT-373: propagating whitespace fix for BTagging changes needed to support the online Hybrid Tagger tuning to 21.2

See merge request atlas/athena!21119

add a basic test for EventLoop grid driver

See merge request atlas/athena!22232

Until now after calling xAOD::TEvent::readFrom(...), the first event (or actually
whatever the internal m_entry variable was set to...) was automatically made
available to the user from the file. This update changes this. So that after
calling readFrom(...), only the metadata of the input file is available.
To get access to the event data, xAOD::TEvent::getEntry(...) must be called
with a valid entry number first.

This is to serve as a test ground for possible EventLoop code updates.

At the same time tried to clean up the package's configuration a bit,
which included removing some unnecessary includes in gt_UnitTestAlgXAOD.cxx.

Updates for JETM12 - track pT cut and ttbar selection

See merge request atlas/athena!22247