fix(triple): clean up REST form request decoder

[juzi050]

What is the purpose of the change?

Closes #16403.

When Triple REST parses an application/x-www-form-urlencoded or multipart/form-data request, DefaultHttpRequest lazily creates and caches an HttpPostRequestDecoder. The decoder was not destroyed after the request completed, so DefaultHttpDataFactory continued to retain the request and its form data.

This change attaches decoder cleanup to the HTTP response lifecycle:

• Add a dedicated method to destroy the decoder cached by DefaultHttpRequest.
• Register decoder cleanup with both HTTP/1.1 and HTTP/2 server response observers.
• Clean up after normal or error response termination, while ensuring cleanup runs only once when multiple terminal signals are received.
• If the transport closes before an asynchronous unary invocation completes normally, defer cleanup until the invocation completes so multipart data remains available to the service method during request processing.

How was it tested?

Extended HttpPostRequestDecoderLifecycleTest to cover the real Triple REST request path:

• Verify that a form request no longer remains tracked by DefaultHttpDataFactory after normal completion.
• Verify decoder cleanup through the HTTP/1.1 response observer.
• Verify that HTTP/2 remote cancellation does not release multipart data while an asynchronous service invocation is still using it, and that the data is released after the invocation completes.
• Verify that custom request adapters that do not use DefaultHttpRequest do not require decoder cleanup.

Added AbstractServerHttpChannelObserverTest to verify that transport closure defers cleanup until normal application completion and that cleanup runs only once when multiple terminal signals are received. The targeted tests passed.

Commands:

./mvnw -pl dubbo-remoting/dubbo-remoting-http12 -Dtest=AbstractServerHttpChannelObserverTest test
./mvnw -pl dubbo-rpc/dubbo-rpc-triple -am -Dtest=HttpPostRequestDecoderLifecycleTest -Dsurefire.failIfNoSpecifiedTests=false test

The changed files were also formatted with codestyle/dubbo_codestyle_for_idea.xml. Maven Checkstyle and Spotless checks passed.

Checklist

• Make sure there is a GitHub issue for the change.
• Write a pull request description that is detailed enough to understand what the pull request does, how it does it, and why it does it.
• Write the necessary unit tests to verify the code.
• Make sure GitHub Actions can pass.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 61.04%. Comparing base (3a30432) to head (b931dc5).

Additional details and impacted files

@@             Coverage Diff              @@
##                3.3   #16412      +/-   ##
============================================
+ Coverage     60.87%   61.04%   +0.16%     
+ Complexity    11766    11765       -1     
============================================
  Files          1953     1953              
  Lines         89273    89304      +31     
  Branches      13473    13478       +5     
============================================
+ Hits          54346    54514     +168     
+ Misses        29333    29183     -150     
- Partials       5594     5607      +13     

Flag	Coverage Δ
integration-tests-java21	32.15% <45.45%> (+0.01%)	⬆️
integration-tests-java8	32.20% <45.45%> (-0.01%)	⬇️
samples-tests-java21	32.13% <0.00%> (-0.07%)	⬇️
samples-tests-java8	29.74% <0.00%> (-0.05%)	⬇️
unit-tests-java11	59.28% <100.00%> (+0.16%)	⬆️
unit-tests-java17	58.76% <100.00%> (+0.12%)	⬆️
unit-tests-java21	58.79% <100.00%> (+0.18%)	⬆️
unit-tests-java25	58.72% <100.00%> (+0.16%)	⬆️
unit-tests-java8	59.30% <100.00%> (+0.15%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ 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.