fix: use DashScope-compatible parameters in Qwen Code provider (#12067)

[roomote-v0]

Related GitHub Issue

Closes: #12067

Description

This PR attempts to address Issue #12067 where Qwen Code provider returns 400 Bad Request errors for all requests. Feedback and guidance are welcome.

The root cause is that the Qwen Code provider sends several OpenAI-specific parameters that DashScope's compatible-mode API does not support. This PR makes the following changes to src/api/providers/qwen-code.ts:

1. max_completion_tokens replaced with max_tokens - DashScope uses the standard max_tokens parameter, not the newer OpenAI-specific max_completion_tokens.
2. Removed parallel_tool_calls - DashScope does not support this OpenAI-specific parameter.
3. Removed stream_options - DashScope does not support stream_options: { include_usage: true }.
4. Overrode convertToolsForOpenAI - DashScope does not support OpenAI's strict: true mode on tool definitions. The override filters for function tools without adding strict mode or schema transformations.

Test Procedure

• Ran cd src && npx vitest run api/providers/__tests__/qwen-code-native-tools.spec.ts - all 12 tests pass.
• Updated existing tests that asserted parallel_tool_calls: true to assert it is NOT present.
• Added new DashScope API Compatibility test suite verifying:
  • max_tokens is used instead of max_completion_tokens
  • stream_options is not included
  • parallel_tool_calls is not included
  • strict: true is not set on tool definitions
  • completePrompt also uses max_tokens

Pre-Submission Checklist

• Issue Linked: This PR is linked to an approved GitHub Issue (see "Related GitHub Issue" above).
• Scope: My changes are focused on the linked issue (one major feature/fix per PR).
• Self-Review: I have performed a thorough self-review of my code.
• Testing: New and/or updated tests have been added to cover my changes.
• Documentation Impact: No documentation updates are required.
• Contribution Guidelines: I have read and agree to the Contributor Guidelines.

Documentation Updates

• No documentation updates are required.

Additional Notes

This fix follows the same pattern used by other OpenAI-compatible providers (like LM Studio and Moonshot) that adapt request parameters to match what their target API actually supports.

Interactively review PR in Roo Code Cloud

[romandesign]

Is there a way to implement this fix right now, or do I need to wait until a new version is released?