Refactor benchmark CI

[XSAM]

Resolves #4537

This PR replaces the github-action-benchmark approach with CodSpeed.

CodSpeed addresses:

• The benchmark conflict issue: Add continuous benchmarks #4537 (comment)
• The maintenance burden for GitHub pages.
• It allows partial performance test, which enables us to run benchmark for PRs in the future

An example of benchmark result: https://codspeed.io/XSAM/opentelemetry-go/runs/697aa351f5662999a88a8bf3

Benchmark page for opentelemetry-go: https://codspeed.io/open-telemetry/opentelemetry-go

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.7%. Comparing base (4ba8200) to head (a1c2448).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@          Coverage Diff          @@
##            main   #7873   +/-   ##
=====================================
  Coverage   81.7%   81.7%           
=====================================
  Files        304     304           
  Lines      23342   23342           
=====================================
+ Hits       19086   19090    +4     
+ Misses      3867    3864    -3     
+ Partials     389     388    -1     

see 2 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[dmathieu]

While I have no objection for this (it fixes the conflict issue which is a big improvement), I want to note that it also adds a dependency on an external private company (we've seen cases where external things like this dropped our access overnight).
Also, on the collector, CodSpeed ends up not being very relevant (see open-telemetry/opentelemetry-collector#14521 for example).

[XSAM]

@dmathieu

Also, on the collector, CodSpeed ends up not being very relevant (see open-telemetry/opentelemetry-collector#14521 for example).

My assumption is because the collector repo uses ubuntu-latest, which is VM where it is a noisy environment, to run benchmarks.

https://github.com/open-telemetry/opentelemetry-collector/blob/5e3bf40de1335e61ab1f9550ccf1d9bcf8a8b58f/.github/workflows/go-benchmarks.yml#L13

We use bare-metal to run benchmarks instead. This would give us a stable result.

[pellared]

@XSAM, bump

[pellared]

Should we add https://codspeed.io/open-telemetry/opentelemetry-go somewhere in README.md?

[XSAM]

Should we add https://codspeed.io/open-telemetry/opentelemetry-go somewhere in README.md?

Yeah, I am planning to do that once we can confirm the benchmark result is stable enough to be useful.