feat: new models added and custom model code entry for all AI providers

[Murtaza-codes]

Summary of Changes

• New Gemini Models Added (src/config/aiProviders.ts):

  • Added latest Gemini models:
    • gemini-2.5-flash (Fast, balanced price/performance)
    • gemini-2.5-pro (Flagship complex reasoning)
    • gemini-2.5-flash-lite (Ultra-low latency default)
    • gemini-2.0-flash (Stable next-generation multimodal)
    • gemini-2.0-flash-lite (High throughput, cost efficient)
    • gemini-2.0-flash-thinking-exp-01-21 (Reasoning & thinking process)
    • gemini-2.0-pro-exp-02-05 (Advanced coding and logic)
    • gemini-1.5-pro & gemini-1.5-flash (Stable legacy models)
    • gemini-3-flash & gemini-3-pro (Preview)

• Custom Model Code Entry for All Providers (src/components/SettingsModal/AISettings.tsx):

  • Allowed manual model ID entry across all providers (Gemini, OpenAI, Claude, Groq, NVIDIA, Cerebras, Mistral, OpenRouter, Ollama, and Custom).
  • Added a toggle button ("Enter model code" / "Select from presets") in the header of the model section.
  • Added an inline "✏️ Enter custom model code..." option at the end of the preset dropdown.
  • Added quick-selection preset pills when in custom mode for fast switching.
  • Preserved custom values when reopening settings.

• Localization & Tests:

  • Synced new translation keys and model metadata across all 7 supported locales.
  • Passed TypeScript type checks (tsc --noEmit).
  • Passed all related unit and integration tests.