[chore]: Add Profiles Marshaler to otlptext

[julianocosta89]

Description

This is the initial work towards #11155.
This PR adds profiles marshalers to otlptext, which will be used in the debugexporter later on.

To continue the work on the debugexporter I'd wait on the PR #11131, as it adds some shared components.

Testing

Added TestProfilesText, testing the following scenarios:

• empty_profiles
• two_profiles

[codecov]

Codecov Report

Attention: Patch coverage is 97.36842% with 4 lines in your changes missing coverage. Please review.

Project coverage is 91.58%. Comparing base (080b765) to head (3393f24).
Report is 4 commits behind head on main.

Files with missing lines	Patch %	Lines
exporter/internal/otlptext/databuffer.go	96.19%	2 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #11161      +/-   ##
==========================================
+ Coverage   91.54%   91.58%   +0.04%     
==========================================
  Files         428      429       +1     
  Lines       20233    20385     +152     
==========================================
+ Hits        18522    18670     +148     
- Misses       1337     1339       +2     
- Partials      374      376       +2     

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

[julianocosta89]

I'm looking for some guidance here.
This test is flaky: https://github.com/open-telemetry/opentelemetry-collector/actions/runs/10957119763/job/30424592592

This is because sometimes we get the returned value as:

-> value: Map({"intValue":"42"})
-> key: Str(answer)

and sometimes like this:

-> key: Str(answer)
-> value: Map({"intValue":"42"})

Any suggestions?

[dmathieu]

Your failing issue is because maps are not ordered in Go. So when the attribute table is being built with FromRaw, the generated slice doesn't guarantee order.

We could possibly ensure order by sorting the keys before injecting in FromRaw. But I'm not sure that's relevant. Maybe figuring out a way to not rely on order of the output would be better?

[julianocosta89]

Your failing issue is because maps are not ordered in Go. So when the attribute table is being built with FromRaw, the generated slice doesn't guarantee order.

We could possibly ensure order by sorting the keys before injecting in FromRaw. But I'm not sure that's relevant. Maybe figuring out a way to not rely on order of the output would be better?

The problem is that the test is a simple comparison created_prof == two_profiles.out.

I've simplified the test to validate the AttributeTable with just one element to avoid the flakiness.

[mx-psi]

Since we don't give any guarantees regarding the debug exporter output, I don't think we need to get it perfect, and this looks good enough to me. Merging!