Add fallback end time for spans in a terminated session

[bidetofevil]

Goal

If a cached session is killed by the user, the endTime of the session object will not be populated. In that case, use the terminationTime, followed by the lastHeartbeatTime, followed by the start time to ensure that the span always has an end time

Testing

Add to unit test

[codecov]

Codecov Report

Attention: Patch coverage is 50.00000% with 2 lines in your changes are missing coverage. Please review.

Project coverage is 80.90%. Comparing base (be623ba) to head (7ca1981).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #884      +/-   ##
==========================================
- Coverage   80.92%   80.90%   -0.02%     
==========================================
  Files         435      435              
  Lines       11616    11613       -3     
  Branches     1764     1764              
==========================================
- Hits         9400     9396       -4     
  Misses       1438     1438              
- Partials      778      779       +1     

Files	Coverage Δ
...mbracesdk/comms/delivery/EmbraceDeliveryService.kt	58.46% <50.00%> (+1.10%)	⬆️

... and 2 files with indirect coverage changes

[fractalwrench]

LGTM pending one comment on the start time/end time potentially being the same for some sessions

[fractalwrench]

This feels like it's going to surprise someone 6 months down the line when they ask why a span has the same start + end time. Would it be better to set a null value, or are there constraints that mean that's not possible?

[bidetofevil]

Hopefully this is not actually needed as terminationTime should always be set if forceQuit is set to true. That said, i added lastHeartbeat and start times just in case.

I agree that start time isn't great, but it's better than 0 (which leads to a negative duration on the backend) and null (ended spans should have an endTime). I'll ping the backend folks when I ping them about why endTime for the session is null for a forceQuit about this as well.