feat(feedback/LLM): update vertex provider to use gemini models

[aliu39]

For gemini models, it's recommended to use the genai SDK instead of posting to a URL like we were for text-bison. We need this before we can request the new model, 2.0 flash lite.

Spam detection has been disabled in all regions except S4S, so we don't need to worry about errors if this is deployed before options, or vice versa

• Depends on pypi: add_pkg google-genai==1.13.0 and upgrade openai==1.77.0 pypi#1419
• Bump in sentry: ref(feedback/llm): add google-genai and bump openai==1.77.0 and pydantic==2.11.4 #91028
• Depends on options: https://github.com/getsentry/getsentry/pull/17378

Ref #90986

[aliu39]

Closing since there's a blocker on adding the genai dependency - see #91028 (comment)

There's a simpler way to directly post to the gemini API using :generateContent instead of :predict - I misinterpreted the error msg pointing me toward the SDK