[fix] Add tool call parser KimiK2RawIdDetector to keep original tool call id.

[guapisolo]

Motivation

The default kimi_k2 path renumbers ids via history_tool_calls_cnt + tool_index in the serving layer so that multi-turn conversations get globally unique, monotonically increasing ids (see PR #10600). That is the right behavior for chat use cases.
However, RL training has the opposite requirement: the trajectory must round-trip the exact tool_call_id the model produced (e.g. functions.foo:5), so that the follow-up tool result turn references the same id the policy emitted. This subclass exists purely as a marker so the serving layer can branch on the parser name and use ToolCallItem.tool_call_id directly. Parsing logic is identical to KimiK2Detector.

Modifications

Add a new KimiK2RawIdDetector to tool call parser, distinguished from original kimi_k2 tool call parser.

Accuracy Tests

Speed Tests and Profiling

Checklist

• Format your code according to the Format code with pre-commit.
• Add unit tests according to the Run and add unit tests.
• Update documentation according to Write documentations.
• Provide accuracy and speed benchmark results according to Test the accuracy and Benchmark the speed.
• Follow the SGLang code style guidance.

Review and Merge Process

1. Ping Merge Oncalls to start the process. See the PR Merge Process.
2. Get approvals from CODEOWNERS and other reviewers.
3. Trigger CI tests with comments or contact authorized users to do so.
   • Common commands include /tag-and-rerun-ci, /tag-run-ci-label, /rerun-failed-ci
4. After green CI and required approvals, ask Merge Oncalls or people with Write permission to merge the PR.

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!