[Feat] Add official GLM 5.3 support to Z.ai API

[zoomote]

​Opened on behalf of @taltas. Follow up by mentioning @roomote, in the web UI, or in Slack.

What changed

• Expose glm-5.3 on the standard international Z.ai API endpoint as well as Coding Plans.
• Apply the official always-on thinking, preserved-thinking, reasoning-effort, and temperature defaults to streaming and non-streaming requests.
• Update provider and model-selection coverage for the official catalog behavior.

Why this change was made

GLM 5.3 is now officially available through Z.ai's Chat Completion API. Zoo Code's preliminary integration was Coding Plan-only and did not send the finalized preserved-thinking parameters on every request path.

Impact

International Z.ai API users can select GLM 5.3 directly. Requests preserve reasoning across turns with clear_thinking: false, reject stale disable settings by keeping thinking enabled, support only low, high, and max effort, and default to the provider-recommended temperature of 1 while retaining explicit user overrides.

Related PRs

• Zoo Code Docs

[codecov]

Codecov Report

❌ Patch coverage is 86.36364% with 3 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/api/providers/zai.ts	86.36%	1 Missing and 2 partials ⚠️

📢 Thoughts on this report? Let us know!