feat: make Gemini model configurable via settings

[srijnabhargav]

Make Gemini model configurable via settings (align with Groq)

Problem

Currently, the Gemini integration uses a hard-coded model in the API request URL.
This limits flexibility and makes it inconsistent with the Groq integration, where the model can be configured via app settings.

As a developer

I need to

Allow configuring the Gemini model through app settings instead of hard-coding it.

So that

Different Gemini models can be used without code changes, and configuration remains consistent across LLM providers.

Acceptance Criteria

• Add a Gemini model input in app settings
• Read the Gemini model from settings at runtime
• Remove hard-coded Gemini model from the API URL