Make IOVSvcTool::handle vaguely re-entrant

This MR is an attempt to fix ATR-22684.

The (unproven) suspicion is that the IOVSvcTool triggers suprious callbacks when the incident handler is concurrently exectued in multiple threads. With this MR, the handle-method gets a bit more reentrant. Full re-entrancy is hard to achieve but at least the sections of the incident-handler that get executed after the first event should be re-entrant. The most important change is to get rid of the m_currTime class member to store the IOVTime of the current event. There are other code-changes, in particular the updates std::map-access methods, that are leftovers from an (failed) attempt to get full const-correctness and re-entrancy.

A bit of background info: We were hoping to get rid of IOV-callbacks alltogether in the MT migration. But at least in the short-term, we have to support IOV-callbacks that are fired only once, in the beginning of the job. The infrastructure has to be sufficiently thread-safe to support this use-case.

cc @leggett, @tsulaia, @emoyse and @elmsheus

This merge request affects 1 package:

• Control/IOVSvc

This merge request affects 6 files:

• Control/IOVSvc/IOVSvc/IIOVSvcTool.h
• Control/IOVSvc/src/CBNode.cxx
• Control/IOVSvc/src/CBNode.h
• Control/IOVSvc/src/IOVSvc.cxx
• Control/IOVSvc/src/IOVSvcTool.cxx
• Control/IOVSvc/src/IOVSvcTool.h

Adding @ssnyder as watcher

added Core master review-pending-level-1 labels

CI Result SUCCESS (hash 3f00615b)

	Athena	AthSimulation	AthGeneration	AnalysisBase	AthAnalysis
externals
cmake
make
required tests
optional tests

Full details available on this CI monitor view
Athena: number of compilation errors 0, warnings 0
AthSimulation: number of compilation errors 0, warnings 0
AthGeneration: number of compilation errors 0, warnings 0
AnalysisBase: number of compilation errors 0, warnings 0
AthAnalysis: number of compilation errors 0, warnings 0
For experts only: Jenkins output [CI-MERGE-REQUEST-CC7 27958]

Thanks @wlampl! This looks great!

If it really fixes the poblem we'll only know with the next large-scale test.

added review-approved label and removed review-pending-level-1 label

merged

mentioned in commit 4111dc38

added sweep:ignore label