ci: split lint into parallel jobs and remove redundant checks

[bryantbiggs]

Summary

The CI lint job is the most frequent source of PR failures. After investigating ~20 recent failures, every single one was a legitimate clippy error (dead code, unused variables, etc.) caught by cargo hack --each-feature — not resource exhaustion or flakiness.

The root cause: PRs introduce code that compiles fine with --all-features but triggers warnings when individual features are tested in isolation (e.g., a function only used with grpc-tonic enabled but missing #[cfg(feature = "grpc-tonic")]). Contributors don't know to run cargo hack locally, so they only discover the issue after pushing.

This PR improves the situation by:

• Splitting lint into two parallel CI jobs for faster feedback:

  • lint: format, workspace clippy, and multi-feature combo checks (~1-2 min)
  • lint-feature-matrix: per-feature clippy via cargo hack --each-feature (~5-6 min)

  Previously all ~150 clippy invocations ran sequentially in a single job. Now the quick checks give immediate feedback while the thorough feature matrix runs in parallel.

• Removing 6 redundant cargo_feature checks from lint.sh that are already covered by cargo hack --each-feature:

  • opentelemetry-otlp "default", "http-proto", "metrics" (single-feature, already tested individually by cargo hack)
  • opentelemetry-jaeger-propagator "default" (same)
  • opentelemetry-proto "default", "full" (same)

• Updating CONTRIBUTING.md to document the cargo hack lint command so contributors can catch per-feature issues locally before pushing.

• Updating scripts/precommit.sh to include the feature matrix check.

Test plan

• Verify lint job passes (format + workspace clippy + multi-feature combos)
• Verify lint-feature-matrix job passes (cargo hack per-feature checks)
• Confirm both jobs run in parallel, not sequentially
• Confirm no regressions in lint coverage (all multi-feature combos still checked)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.8%. Comparing base (ec289cb) to head (8941537).

Additional details and impacted files

@@          Coverage Diff          @@
##            main   #3366   +/-   ##
=====================================
  Coverage   82.8%   82.8%           
=====================================
  Files        130     130           
  Lines      27289   27289           
=====================================
  Hits       22622   22622           
  Misses      4667    4667           

☔ 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.

[scottgerring]

I think we need --all-targets ?

Suggested change

	cargo hack --each-feature --no-dev-deps clippy -- -Dwarnings
	cargo hack --each-feature --no-dev-deps clippy --all-targets -- -Dwarnings

cargo-hack still checks each feature combination, but unlike cargo_feature calls it does not pass --all-targets, so clippy coverage for tests/examples/benches under those feature combinations is reduced.