FIX: add latest gemini models to ModelProviderMapping (Issue 188)

[ShubhanC]

What

Fixes #188. Gemini model support isn't updated to the latest models. Some are completely unsupported by Google and new ones have been added that can be used.

Fix

Made changes to provider.json that removed discontinued models and added new ones.

Removed the following models:

1. Gemini 2.5 Pro (gemini-2.5-pro)
2. Gemini 2.0 Flash (gemini-2.0-flash)
3. Gemini 2.0 Flash Lite (gemini-2.0-flash-lite)

Added the following models:

1. Gemini 3.6 Flash (gemini-3.6-flash)
2. Gemini 3.5 Flash (gemini-3.5-flash)
3. Gemini 3.5 Flash Lite (gemini-3.5-flash-lite)
4. Gemma 4 26B (gemma-4-26b-a4b-it)
5. Gemma 4 31B (gemma-4-31b-it)

Parameters for all new models are:
{ "temperature": 0.1, "top_p": 0.9 }

For Gemma 4 26B, the parameters also include "repeat_penalty": 1.2 to ensure the model finishes in sufficient time.

Tests

I performed simple tests for all models (including existing models) on both the given sample resumé and my own personal resumé. Here are the results:

O --> works
X --> fails

First letter is sample resumé, second is personal resumé

gemini-2.5-flash: OO
gemini-2.5-flash-lite: OO
gemini-3.6-flash: OO
gemini-3.5-flash: OO
gemini-3.5-flash-lite: OO
gemma-4-26b-a4b-it: OO
gemma-4-31b-it: OO

Hopefully this allows for future updating of Google gemini models.

[ShubhanC]

fixes #188