DM-45340: Improve exception handling in quantum executors

[andy-slac]

This should solve an issue with exceptions happening at unpredictable locations in a SingleQuantumExecutor class. MPGraphExecutor in a single-process mode should also behave reasonably for random exceptions. Multi-process mode has more complex internal state. If a client of the code catches an exception and tries to reuse the executor, the state may be inconsistent, depending on where the exception happens.

Checklist

• ran Jenkins
• added a release note for user-visible changes to doc/changes

[codecov]

Codecov Report

Attention: Patch coverage is 88.57143% with 12 lines in your changes missing coverage. Please review.

Project coverage is 87.28%. Comparing base (efd0d94) to head (cb84fe4).

Files	Patch %	Lines
tests/test_executors.py	90.36%	3 Missing and 5 partials ⚠️
python/lsst/ctrl/mpexec/mpGraphExecutor.py	76.47%	2 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #297      +/-   ##
==========================================
+ Coverage   87.25%   87.28%   +0.03%     
==========================================
  Files          50       50              
  Lines        4581     4585       +4     
  Branches      788      788              
==========================================
+ Hits         3997     4002       +5     
  Misses        422      422              
+ Partials      162      161       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[kfindeisen]

This PR doesn't actually fix the problem -- the new code in MpGraphExecutor is exactly as vulnerable to interrupts as the old code in SingleQuantumExecutor was (raising in the first task instead of a later one was a red herring). Having now seen the code for myself, I suspect catching the exception from getReport is the best solution.

I also recommend doing updates, code style, and maintenance on separate commits from the main work; it usually makes sense to do them first.

[andy-slac]

@kfindeisen, I am certain that this indeed fixes the problem. Can you reproduce the crash that you saw before?

[kfindeisen]

Looks much better, thanks.