introduce pprofile.AddAttribute helper

[dmathieu]

Description

This introduces the pprofile.AddAttribute helper method so profile extensions can modify attributes.
Closes #12206

BenchmarkAddAttribute
BenchmarkAddAttribute/with_a_new_string_attribute
BenchmarkAddAttribute/with_a_new_string_attribute-10            30461164                39.55 ns/op           32 B/op          2 allocs/op
BenchmarkAddAttribute/with_an_existing_attribute
BenchmarkAddAttribute/with_an_existing_attribute-10             30714298                39.44 ns/op           32 B/op          2 allocs/op
BenchmarkAddAttribute/with_a_duplicate_attribute
BenchmarkAddAttribute/with_a_duplicate_attribute-10             29132547                39.47 ns/op           32 B/op          2 allocs/op
BenchmarkAddAttribute/with_a_hundred_attributes_to_loop_through
BenchmarkAddAttribute/with_a_hundred_attributes_to_loop_through-10               6762793               176.8 ns/op            32 B/op          2 allocs/op

Those benchmarks kind of have an N+1 issue, where the method will take longer the more attributes there are.
However, since the attribute table is a map, we can't go around with looping through it.

I also had to use FromRaw and AsRaw in there, to properly handle any values.

This currently doesn't support removing entries. But doing so would also have a non-trivial performance impact, since we'd have to loop through the AttributeIndices slice.

[codecov]

Codecov Report

Attention: Patch coverage is 66.66667% with 9 lines in your changes missing coverage. Please review.

Project coverage is 92.07%. Comparing base (dae5d19) to head (8e2ff1b).
Report is 12 commits behind head on main.

Files with missing lines	Patch %	Lines
pdata/pprofile/attributes.go	66.66%	6 Missing and 3 partials ⚠️

❌ Your patch check has failed because the patch coverage (66.66%) is below the target coverage (95.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #12390      +/-   ##
==========================================
- Coverage   92.19%   92.07%   -0.13%     
==========================================
  Files         465      465              
  Lines       25285    25200      -85     
==========================================
- Hits        23311    23202     -109     
- Misses       1575     1598      +23     
- Partials      399      400       +1     

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

[dmathieu]

@bogdandrutu I'm happy to hear your input on improvements here, especially with the loop through all attributes issue, and the use of FromRaw and AsRaw to be able to have an any value.

[jmacd]

How come the attributable type is not an exported type (of some package)?

[dmathieu]

attributable is there to allow the behavior on any of the generated types that have attribute indices.
Location, Mapping, Sample.
Folks will only ever need to manipulate the generated struct. They don't need the interface.