Stop ignoring long measurements in HistogramExemplarReservoir

[jack-berg]

As noted by @dashpole here there's a bug in the HistogramExemplarReservoir implementation.

HistogramPointData only supports double exemplars. Despite our API supporting both LongHistogram and DoubleHistogram, long values are simply converted to doubles in the aggregator. However, the HistogramExemplarReservoir didn't get the memo and reports a value of 0.0 when any long measurement is recorded as an exemplar.

This fixes the issue by casting long measurements to double before storing in the reservoir.

[codecov]

Codecov Report

Base: 91.08% // Head: 91.09% // Increases project coverage by +0.00% 🎉

Coverage data is based on head (d0d9abb) compared to base (f183114).
Patch coverage: 100.00% of modified lines in pull request are covered.

Additional details and impacted files

@@            Coverage Diff            @@
##               main    #5216   +/-   ##
=========================================
  Coverage     91.08%   91.09%           
  Complexity     4875     4875           
=========================================
  Files           549      549           
  Lines         14418    14420    +2     
  Branches       1372     1372           
=========================================
+ Hits          13133    13136    +3     
  Misses          891      891           
+ Partials        394      393    -1     

Impacted Files	Coverage Δ
.../internal/exemplar/HistogramExemplarReservoir.java	90.90% <100.00%> (+2.02%)	⬆️
...ava/io/opentelemetry/sdk/internal/RateLimiter.java	94.11% <0.00%> (-5.89%)	⬇️
...telemetry/sdk/trace/export/BatchSpanProcessor.java	94.48% <0.00%> (+0.68%)	⬆️
...metry/sdk/logs/export/BatchLogRecordProcessor.java	91.48% <0.00%> (+0.70%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.