Update Prometheus client to 1.4.3

[zeitlinger]

Summary

Simplified replacement for #7588 - updates the Prometheus client from 1.3.10 to 1.4.3 without adding a UTF-8 support flag.

In Prometheus client 1.4.x, sanitizeMetricName and sanitizeLabelName no longer convert non-standard characters (dots, dashes, non-ASCII) to underscores, since the client now supports UTF-8 natively. To maintain backward-compatible behavior, prometheusName() is called first to convert names to legacy Prometheus format before sanitization.

Changes:

• Bump prometheusServerVersion from 1.3.10 to 1.4.3
• Add prometheusName() calls for metric names and label names to preserve legacy escaping
• Update shaded protobuf import in test (4.31.1 → 4.33.0)

Test plan

• All existing prometheus exporter tests pass (129 tests)
• MetricReaderFactoryTest in sdk-extensions/incubator passes

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.20%. Comparing base (7ce6c5c) to head (2b8cac6).

Additional details and impacted files

@@             Coverage Diff              @@
##               main    #8080      +/-   ##
============================================
- Coverage     90.20%   90.20%   -0.01%     
  Complexity     7595     7595              
============================================
  Files           841      841              
  Lines         22919    22920       +1     
  Branches       2291     2291              
============================================
  Hits          20674    20674              
  Misses         1529     1529              
- Partials        716      717       +1     

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

🚀 New features to boost your workflow:

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

[jack-berg]

@zeitlinger when I look into the code I see prometheusName calls escapeName(String, EscapingScheme), where EscapingScheme is an enum with values: ALLOW_UTF8, UNDERSCORE_ESCAPING, DOTS_ESCAPING, VALUE_ENCODING_ESCAPING

And I've seen some hints in the code that the metric and attribute name escaping is influenced by headers requested by the client. Do we need to escape at all or will prometheus do we want by default?

Also, we've had offline conversations about the prometheus client library automatically appending suffixes like _total and units, and not providing an opt out mechanism, which makes it problematic for us in opentelemetry-java to implement the translation strategy option. Want to confirm this is still the case with this new version. Seems a bit strange that prometheus would have facilities to produce a full unescaped utf8 metric name but still always produces metric names with the suffixes.