feat: add chat decorator to chat function to expose it as an API endpoint

refactor: migrate model inference from local Transformers to OpenRouter API with native reasoning support

feat: update GPU decorator to xlarge size for chat function

chore: update model_id to Laguna-XS.2 base version

feat: enable trust_remote_code for tokenizer and model initialization

chore: update model_id to Laguna-XS.2-FP8 for optimized inference

feat: initialize Laguna-XS.2 project with Gradio backend and custom HTML chat interface