Add a benchmark for histogram allocations

[MadVikingGod]

This adds a simple histogram benchmark to measure the allocations done in collect().

The output from the benchmark is currently:

BenchmarkCollectHistograms/1-16          1166540              1023 ns/op             592 B/op          6 allocs/op
BenchmarkCollectHistograms/5-16           302680              3710 ns/op            2640 B/op         22 allocs/op
BenchmarkCollectHistograms/10-16          160407              6953 ns/op            5200 B/op         42 allocs/op
BenchmarkCollectHistograms/25-16           66810             17389 ns/op           13072 B/op        102 allocs/op

So from a linear regression we can see there is a baseline of 2 allocations of any number of histograms, from creating rm.ScopeMetrics and rm.ScopeMetrics.Metrics. ¹
We can also see the slope of the regression is 4, so there are 4 allocations per histogram created. 2 of these are created by copying the bounds and the count slices. ²

This is just something to set our baseline for #3047.

Footnotes

1. Not tested here, but we would expect this to be 1+ the number of scopes that produced metrics. ↩

2. This is also done for Delta Histograms. ↩

[codecov]

Codecov Report

Merging #3635 (8dbede3) into main (aa51224) will increase coverage by 0.0%.
The diff coverage is n/a.

Additional details and impacted files

@@          Coverage Diff          @@
##            main   #3635   +/-   ##
=====================================
  Coverage   79.6%   79.7%           
=====================================
  Files        171     171           
  Lines      12673   12673           
=====================================
+ Hits       10099   10105    +6     
+ Misses      2361    2355    -6     
  Partials     213     213           

Impacted Files	Coverage Δ
exporters/jaeger/jaeger.go	91.1% <0.0%> (+0.8%)	⬆️
sdk/trace/batch_span_processor.go	81.9% <0.0%> (+1.7%)	⬆️