fix(model-requirements): use supported variant for gemini-3-pro

[sk0x0y]

Summary

Fixes gemini-3-pro fallback entries that use unsupported variant: "max".

According to Google's official Gemini API documentation, Gemini 3 Pro only supports "low" and "high" thinking levels:

thinkingLevel (string) - Controls the model's reasoning depth.

• Gemini 3 Pro: "low", "high" (default: "high")
• Gemini 3 Flash: "minimal", "low", "medium", "high" (default: "high")

Using "max" results in API errors (400 Bad Request) when the fallback resolves to Gemini 3 Pro.

Changes

Changed variant: "max" → variant: "high" for gemini-3-pro in:

Type	Name
Agent	oracle
Agent	metis
Agent	momus
Category	ultrabrain
Category	deep
Category	artistry

Notes

• "high" is already the maximum thinking level for Gemini 3 Pro
• Other models (claude-opus-4-5, gpt-5.2) correctly use their respective max variants
• No behavioral change since "high" = maximum reasoning for this model

Closes #1433

---

Summary by cubic

Replaced gemini-3-pro fallback variant "max" with "high" to match Google’s supported thinking levels. This prevents 400 errors when fallbacks resolve to Gemini 3 Pro.

• Bug Fixes
  • Updated gemini-3-pro variant to "high" in oracle, metis, momus agents and ultrabrain, deep, artistry categories.

^{Written for commit dfcfcfa. Summary will update on new commits.}

[github-actions]

All contributors have signed the CLA. Thank you! ✅
_{Posted by the CLA Assistant Lite bot.}

[cubic-dev-ai]

No issues found across 1 file

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

[sk0x0y]

I have read the CLA Document and I hereby sign the CLA

[sk0x0y]

recheck