fix: match rate limits by data category not envelope item type

[jamescrosswell]

Summary

Rate limits in the X-Sentry-Rate-Limits header are keyed by data category (error, monitor, log_item, metric_bucket, …), which is not always the same as the envelope item type (event, check_in, log, statsd).

RateLimitCategory.Matches compared the server's category name against the raw envelope item type, so category-specific limits whose name differs from the item type were silently ignored:

Item type	Category Relay sends	Old Matches result
event	error	❌ never matched
check_in	monitor	❌ never matched
log	log_item	❌ never matched

The practical effect: when Sentry rate-limited errors, the SDK didn't recognise the limit, kept sending, got 429d downstream, and client reports mis-attributed the drops to default instead of the real category. This is the concern raised in #4535 — mishandling categories can cause the most important data (errors) to be dropped rather than cleanly backed off.

Changes

• RateLimitCategory.Matches now compares the limit against the item's data category rather than the raw item type — so error / monitor / log_item limits are honoured. The special-case for statsd → metric_bucket is no longer needed.
• EnvelopeItem.DataCategory maps every emitted item type to its correct category: check_in → monitor, log → log_item, trace_metric → trace_metric, statsd → metric_bucket. Client reports (client_report) now map to internal so they are never dropped by a default rate limit.
• BatchProcessor records dropped logs/metrics under their real category (log_item / trace_metric) instead of default.
• DataCategory gains the missing monitor / log_item / trace_metric / metric_bucket constants and the doc link is updated.
• Tests updated to use the real category names (error, not event) and to guard against the regression (item-type strings must not match).

Verified: full Sentry.Tests suite passes on net9.0 (2465 passed, 4 environment-skips).

Closes #4535

[jamescrosswell]

Keep this concise...

Suggested change

	// Rate limits are keyed by data category (e.g. "error", "monitor", "metric_bucket"), which is not
	// always the same as the envelope item type (e.g. "event", "check_in", "statsd"). Compare against the
	// item's data category so limits for those categories are honoured rather than silently ignored.
	// Rate limits are keyed by data category (e.g. "error", "monitor", "metric_bucket")

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 74.63%. Comparing base (4216c80) to head (4d97258).

Additional details and impacted files

@@                     Coverage Diff                     @@
##           fix/5406-pin-xcode-26.6    #5412      +/-   ##
===========================================================
+ Coverage                    74.60%   74.63%   +0.03%     
===========================================================
  Files                          512      512              
  Lines                        18666    18674       +8     
  Branches                      3659     3657       -2     
===========================================================
+ Hits                         13925    13937      +12     
+ Misses                        3867     3864       -3     
+ Partials                       874      873       -1     

☔ View full report in Codecov by Harness.
📢 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.

[jamescrosswell]

This could be simplified to a lambda that simply returns IsMatchAll || string.Equals(...