Use event-local memory pool when reading RawEvent

This uses the tools of !2270 (merged) when reading raw events from MDF.

There is a slight complication thanks to a bug in ROOT: https://sft.its.cern.ch/jira/browse/ROOT-10526 A fix has been proposed, but I haven't been able to test it yet. The fix should make it to the 6.20 release, and presumably will be available via the LCG dev3 and/or dev4 slots before then. The effect is that ROOT I/O of LHCb::RawEvent is not possible with this MR and without the ROOT fix.

I added a unit test to this MR that exposes the problem (it segfaults/fails).

cc: @sponce @frankm

mentioned in merge request !2313 (closed)

mentioned in merge request !2270 (merged)

mentioned in issue #73 (closed)

@frankm @sponce could one of you please check the diff and hit "Approve" once you're happy?

added Persistency enhancement labels

assigned to @frankm and @sponce

approved this merge request

Very nice. I should resurrect the MEPs code and merge it with that to see what we gain !

Why not (also) update LHCb::RawBank* allocateBank( size_t len ) (see here, called from createBank) to use the allocator (instead of plain new)?

added lhcb-head label

removed lhcb-head label

mentioned in issue Moore#131 (closed)

https://sft.its.cern.ch/jira/browse/ROOT-10526 is now closed, apparently the fix should be in 6.20/00, 6.18/06. I think this means it will be picked up in both dev3 and dev4 slots, so we could test this MR in one or both of those if we wanted to.

added 35 commits

• 2320dbef...1f102fd8 - 33 commits from branch olupton_custom_allocators
• edfc701f - Migrate DAQ algorithms to use custom memory pool.
• cb204c11 - RawEvent unit test that exposes ROOT-10526.

Compare with previous version

added 18 commits

• cb204c11...25e5092d - 16 commits from branch olupton_custom_allocators
• 2bdb312a - Migrate DAQ algorithms to use custom memory pool.
• f9539121 - RawEvent unit test that exposes ROOT-10526.

Compare with previous version

added needs Gaudi release label

changed the description

changed target branch from olupton_custom_allocators to master

added 20 commits

• f9539121...c5a8d544 - 18 commits from branch master
• 6c8ddb4d - Migrate DAQ algorithms to use custom memory pool.
• 5f6df87b - RawEvent unit test that exposes ROOT-10526.

Compare with previous version

If the dev4 slot was "free" then I think it would be worth testing this.

(I haven't reminded myself how to use dev4 locally yet, so I am yet to test the upstream ROOT fix this depends on)