Fix persistent Connection Errors for OpenAI/DeepSeek: disable HTTP/2 and use robust httpx client

Fix Connection error for OpenAI and DeepSeek: add retry logic and increase timeout to 120s

Improve DeepSeek compatibility (R1/reasoner support) and fix lint errors in main.py

Fix OpenAI gpt-4.1 API parameters: role 'developer', max_completion_tokens, and reasoning effort support; update max_tokens in default config

Increase max_tokens for all models in generation settings to enhance response capacity

Increase max_tokens for generation settings and enhance JSON extraction to handle various markdown formats

Add JSON extraction utility and enhance analysis methods for better error handling

Add placeholders for court decision text and comment in legal position prompt

Fix input detection: prioritize URL input even when method state is incorrect

Fix validation logic: allow processing when URL is provided, show error only when empty

Fix input validation: add specific error messages for empty URL and text inputs

Clean deployment without large index files