appendCAtoAthena: Recusive conversion of GaudiConfig-ListHelper to list, do not append duplicates to SvcMgr, ToolSvc, etc

Reverse use of context in JetBTaggingAlg for JFVtx and SecVtx container (crash with ctx)

Need to flush output in order to get consistent output ordering
between py2 and py3.

Added more Run3 config for MuonCombinedReconstruction, added stack trace printing in case of unmerged CA

StatusCodeSvc is deprecated. Remove all references to it from job
options.

Edward Moyse authored 5 years ago and Adam Edward Barton committed 5 years ago

Another interim commit. Some problems spotted, notably that the RungeKutta needs to have a CA configuration - so here just using bare configurable.

Flush stdout at the start of ComponentAccumulator.run, to make sure all python
output has been printed before we start getting output from Gaudi.
Otherwise, the observed output ordering may differ between py2 and py3.

Scott Snyder authored 5 years ago and Walter Lampl committed 5 years ago

Add option to printConfig to allow restricting the set of components
to be printed.

Sometimes, i have observed jobs using new configuration failing on a call to Gaudi
with the error that gbl.Gaudi is not a Gaudi instance.  This can be avoided
by touching the pyroot object earlier.

Remove ForceRead

Tomasz Bold authored 5 years ago and Vakhtang Tsulaia committed 5 years ago

made sure that the top main servvices CA is never merged to another accumulator but the others are merged to it

Edward Moyse authored 5 years ago and Frank Winklmeier committed 5 years ago

Fix handling of single private tools, and reverse order of primary component determination, to favour private tools over services.

This latter change is because you can get a service from another CA Cfg method, but if you explicitly set a private tool, that must be the intended primary component (since private tools cannot be left dangling).