Add account and container metrics for client aborts

[jitheshtr]

Summary

Ambry needs a per-account/container signal for requests that the remote client abandons. This change adds additive abort counters for remote inactivity and failed response writes without changing response status codes or normalizing existing status-metric ordering.

What changed

1. Added ClientAbortCount for all safely attributed client terminations and ServerErrorClientAbortCount only when the abort caused the request's recorded 5xx classification. Both counters are emitted at container and account level, allowing the subset to discount client-caused server errors without hiding a genuine 5xx whose response write later failed.
2. Replaced separate termination guards with one request-scoped atomic outcome shared by idle timeout, channel inactivity, and response-write failure. The first cause wins, so one request contributes to at most one termination-duration histogram and idle cleanup cannot become a client abort.
3. Captured server provenance at the Netty boundary. Outbound server closure sets a channel marker, while the channel also retains the existing RestServerState; service-down shutdown closes are excluded even when Netty bypasses outbound handlers.
4. Covered actual response-write failures, inactive writes, and 100-Continue failures while excluding synthetic cleanup callbacks and lifecycle-only RequestChannelClosed classification. A failed write of an existing 5xx records the general abort and duration but stays out of the subtractable subset.

Suggested reading order: PublicAccessLogHandler, NettyMessageProcessor, NettyResponseChannel, RestRequestMetricsTracker, then metric registration and tests.

Known gaps

Attribution starts only after AccountAndContainerInjector supplies container metrics. Errors such as RequestChannelClosed or router ChannelClosed remain excluded unless a Netty boundary independently proves remote termination. The change adds two series to each existing account/container-operation and account-operation tuple, increasing the per-tuple entity-operation series from 11 to 13.

For bodiless non-GET responses, metadata completion can precede any observable disconnect, so the causal 5xx subset can legitimately remain sparse even when the general abort counter captures later response-write terminations.

Risk Assessment

This is metrics and termination-attribution logic only. It does not modify storage, named-blob operations, TTL, deletion, retries, response codes, response bodies, or client-visible timing. The channel evidence is terminal and channel-local; the termination outcome is request-local and atomic. Negative tests cover idle timeout during streaming, outbound server close, unsafe service shutdown, genuine 5xx response-write failure, lifecycle-only exceptions, inline and deferred inactive-write ordering, and duplicate termination paths. Durability risk: none.

Testing Done

• Local code review completed

Command	Result
Focused API and REST test classes	71 passed
./gradlew :ambry-api:test :ambry-rest:test	API: 235 passed, 6 skipped; REST: 115 passed
./gradlew :ambry-api:javadoc :ambry-rest:javadoc	Passed
Targeted regression mutations	Streamed-content attribution, synthetic-cleanup exclusion, and non-client cause filtering each failed under its matching mutation

---

Tests generated with unit-tests plugin

🤖 Generated with GitHub Copilot CLI

[codecov-commenter]

Codecov Report

❌ Patch coverage is 27.71084% with 60 lines in your changes missing coverage. Please review.
✅ Project coverage is 50.70%. Comparing base (52ba813) to head (3ca9d3e).
⚠️ Report is 416 commits behind head on master.

Files with missing lines	Patch %	Lines
...va/com/github/ambry/rest/NettyResponseChannel.java	0.00%	21 Missing ⚠️
...a/com/github/ambry/rest/NettyMessageProcessor.java	0.00%	18 Missing ⚠️
.../com/github/ambry/rest/PublicAccessLogHandler.java	63.15%	6 Missing and 1 partial ⚠️
...m/github/ambry/rest/RestRequestMetricsTracker.java	53.84%	5 Missing and 1 partial ⚠️
...va/com/github/ambry/frontend/ContainerMetrics.java	0.00%	4 Missing ⚠️
.../github/ambry/frontend/EntityOperationMetrics.java	42.85%	4 Missing ⚠️

Additional details and impacted files

@@              Coverage Diff              @@
##             master    #3296       +/-   ##
=============================================
- Coverage     64.24%   50.70%   -13.54%     
+ Complexity    10398     8689     -1709     
=============================================
  Files           840      938       +98     
  Lines         71755    80652     +8897     
  Branches       8611     9722     +1111     
=============================================
- Hits          46099    40898     -5201     
- Misses        23004    36350    +13346     
- Partials       2652     3404      +752     

☔ 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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.