Measure metrics latency

[fraillt]

Tests how much time it takes to call .add(value, attribs) under various conditions.
It measures several things: average time, median time (p50), p95 and p99.
The way it works is pretty simple:

• call .add(value, attribs) on several threads in the loop and measure each call
• in main thread, keep collecting data (delta collect) with some delay between calls.
  Here's the results on my machine on main branch.

See results.

*** updates, using 4 threads ***
no attribs
	iter 26.43M
	avg 103.0ns
	p50 57.00ns
	p95 177.0ns
	p99 182.0ns
1 attrib
	iter 8.267M
	avg 422.0ns
	p50 362.0ns
	p95 743.0ns
	p99 1.017μs
9 attribs
	iter 4.919M
	avg 718.0ns
	p50 615.0ns
	p95 854.0ns
	p99 1.026μs
*** inserts, using 4 threads ***
1 attrib
	iter 435.1K
	avg 8.931μs
	p50 958.0ns
	p95 64.16μs
	p99 113.8μs
10 attribs
	iter 119.0K
	avg 33.03μs
	p50 4.959μs
	p95 133.9μs
	p99 205.0μs
*** mix mostly updates (100 attribute-sets), using 4 threads ***
10 attribs
	iter 379.1K
	avg 10.36μs
	p50 1.729μs
	p95 67.73μs
	p99 125.9μs
*** updates, using 36 threads ***
no attribs
	iter 40.32M
	avg 682.0ns
	p50 215.0ns
	p95 228.0ns
	p99 243.0ns
1 attrib
	iter 11.72M
	avg 2.991μs
	p50 797.0ns
	p95 1.539μs
	p99 2.264μs
9 attribs
	iter 8.819M
	avg 3.952μs
	p50 933.0ns
	p95 1.442μs
	p99 1.797μs
*** inserts, using 36 threads ***
1 attrib
	iter 279.6K
	avg 128.2μs
	p50 2.757μs
	p95 792.6μs
	p99 1.596ms
10 attribs
	iter 80.01K
	avg 448.5μs
	p50 130.5μs
	p95 1.858ms
	p99 3.044ms
*** mix mostly updates (100 attribute-sets), using 36 threads ***
10 attribs
	iter 1.249M
	avg 28.57μs
	p50 945.0ns
	p95 13.10μs
	p99 871.9μs

Merge requirement checklist

• CONTRIBUTING guidelines followed
• Unit tests added/updated (if applicable)
• Appropriate CHANGELOG.md files updated for non-trivial, user-facing changes
• Changes in public API reviewed (if applicable)

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 79.5%. Comparing base (465fcc2) to head (e0df038).

Additional details and impacted files

@@          Coverage Diff          @@
##            main   #2323   +/-   ##
=====================================
  Coverage   79.5%   79.5%           
=====================================
  Files        123     123           
  Lines      21258   21258           
=====================================
+ Hits       16905   16907    +2     
+ Misses      4353    4351    -2     

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

---

🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests

[cijothomas]

trying to understand what this is simulating? Metric collection interval is by default 30 sec or 60 sec in the spec. What is the scenario for collecting every 500 microsec?