feat: activate thinking mode for Kimi K2.5 on Moonshot AI provider

[roomote]

This PR attempts to address Issue #11033 by enabling thinking mode for the Kimi K2.5 model on the Moonshot AI provider.

Changes

• Added preserveReasoning: true to the kimi-k2.5 model configuration in packages/types/src/providers/moonshot.ts

Why

The kimi-k2.5 model supports reasoning/thinking mode, similar to kimi-k2-thinking. This change enables the thinking mode functionality by setting preserveReasoning: true, which is the same property used by other thinking-enabled models like kimi-k2-thinking in the Moonshot provider and Fireworks provider.

Testing

• All 19 existing moonshot tests pass
• TypeScript type checking passes

Feedback and guidance are welcome.

---

Important

Enables thinking mode for kimi-k2.5 by adding preserveReasoning: true in moonshot.ts, aligning it with other reasoning-enabled models.

• Behavior:
  • Enables thinking mode for kimi-k2.5 by adding preserveReasoning: true in moonshot.ts.
  • Aligns kimi-k2.5 with other reasoning-enabled models like kimi-k2-thinking.
• Testing:
  • All 19 existing Moonshot tests pass.
  • TypeScript type checking passes.

^{This description was created by for 912fa1f. You can customize this summary. It will automatically update as commits are pushed.}

[roomote]

Rooviewer   See task on Roo Cloud

Review complete. No issues found.

The change correctly enables thinking mode for the kimi-k2.5 model by adding preserveReasoning: true, following the same pattern used by kimi-k2-thinking and other reasoning-enabled models across the codebase.

_{Mention @roomote in a comment to request specific changes to this pull request or fix all unresolved issues.}