fix(obs): remove stale replication metric TODOs

[overtrue]

Summary

• Remove stale bucket replication proxy metric TODO comments after verifying the metrics already exist.
• Add regression assertions that DELETE tagging proxied total and failure metrics are emitted by the bucket replication collector.
• Do not change runtime collection logic or request paths.

Backlog: rustfs/backlog#646

Validation

• RED/coverage check: PATH="$HOME/.cargo/bin:$PATH" cargo test -p rustfs-obs test_collect_bucket_replication_metrics -- --nocapture
• PATH="$HOME/.cargo/bin:$PATH" cargo fmt --all --check
• PATH="$HOME/.cargo/bin:$PATH" cargo clippy -p rustfs-obs --tests -- -D warnings

Performance

• No runtime logic changed. The only production change removes stale comments; test-only assertions validate existing metrics emission.

[github-actions]

CLA requirements are satisfied for this pull request.

[Copilot]

Pull request overview

This PR cleans up bucket replication proxy metric schema documentation by removing stale TODO comments and strengthens regression coverage by asserting that DELETE tagging proxied total and failure metrics are emitted by the bucket replication metrics collector.

Changes:

• Removed outdated TODO comments in the bucket replication metric schema where the referenced metrics already exist.
• Updated the bucket replication collector unit test to assert emission (name/labels/value) of DELETE tagging proxied total and failures metrics.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
crates/obs/src/metrics/schema/bucket_replication.rs	Removes stale TODO comments now that the relevant proxied tagging metrics are already defined.
crates/obs/src/metrics/collectors/bucket_replication.rs	Adds/adjusts unit test assertions to validate DELETE tagging proxied total/failure metrics are emitted.