feat(instrumentation-aws-sdk): add gen ai metrics for bedrock

[anuraaga]

Which problem is this PR solving?

• Gen AI defines metrics that should be applied for operations like bedrock runtime

Short description of the changes

• Allows propagating meter to service extensions
• Passes start time to service extensions to allow them to compute operation duration as needed
• Implement metrics for bedrock service extension
• Refactor tests to share a single instrumentation instance

/cc @trentm @codefromthecrypt

[anuraaga]

I tried many attempts to have the meter-enabled instrumentation running for bedrock tests, including stuff like beforeEach(() => registerInstrumentationForTesting(); instrumentation.enable()) but nothing worked, so I ended up refactoring to make sure there's only a single instrumentation instance among all these tests (since mocha doesn't isolate tests). Let me know if there's any better approach

[trentm]

Yah, mocha running all tests in the same single process is a pain, IMHO. (Test frameworks often go this way though because speed.)

Could you add a block comment at the top of the file (after the license block) that explains basically what you've said above.

[codecov]

Codecov Report

Attention: Patch coverage is 96.55172% with 1 line in your changes missing coverage. Please review.

Project coverage is 89.51%. Comparing base (9d84216) to head (00e7ac6).
Report is 2 commits behind head on main.

Files with missing lines	Patch %	Lines
...entelemetry-instrumentation-aws-sdk/src/aws-sdk.ts	87.50%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2771      +/-   ##
==========================================
- Coverage   90.55%   89.51%   -1.05%     
==========================================
  Files         168      172       +4     
  Lines        8076     8332     +256     
  Branches     1547     1593      +46     
==========================================
+ Hits         7313     7458     +145     
- Misses        763      874     +111     

Files with missing lines	Coverage Δ
...entelemetry-instrumentation-aws-sdk/src/semconv.ts	100.00% <100.00%> (ø)
...ntation-aws-sdk/src/services/ServicesExtensions.ts	96.87% <100.00%> (+0.32%)	⬆️
...umentation-aws-sdk/src/services/bedrock-runtime.ts	98.11% <100.00%> (+0.43%)	⬆️
...entelemetry-instrumentation-aws-sdk/src/aws-sdk.ts	92.21% <87.50%> (-0.34%)	⬇️

... and 6 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[trentm]

nit: A comment linking to https://opentelemetry.io/docs/specs/semconv/gen-ai/gen-ai-metrics/#metric-gen_aiclienttokenusage could be helpful. Ditto for the other metric below.

[trentm]

export const GEN_AI_TOKEN_TYPE_VALUE_COMPLETION = "output" is weird: "COMPLETION" rather than "OUTPUT".
Ah, looks like that'll be fixed by https://github.com/open-telemetry/semantic-conventions/pull/1757/files#diff-0936f2641601c5b370b10dfb33c8f82d6d731a2663cf042eef2aa5114c720902R177
which is probably in semconv v1.31.0, for which the OTel JS semconv package doesn't have a release yet.
So eventually this'll be GEN_AI_TOKEN_TYPE_VALUE_OUTPUT.

[trentm]

Yah, mocha running all tests in the same single process is a pain, IMHO. (Test frameworks often go this way though because speed.)

Could you add a block comment at the top of the file (after the license block) that explains basically what you've said above.

[trentm]

This import from contrib-test-utils doesn't absolutely need to be before ./load-instrumentation does it?

[trentm]

This method is called in InstrumentationAbstract's constructor. Is it necessary to call it again?
The method is also called in initMeterProvider (that calls instrumentation.setMeterProvider(meterProvider);, which I think calls instr._updateMetricInstruments()).