Fix textarea scrollbar arrows showing by default in chat input

Fix post-deployment issues: admin iframe login, false positives, responsive UI, chat memory, stop generation button, stream cut-offs, and alert retries

Add debug logging to _get_llm_endpoint and inner exception details to stream error handlers

Log response body for non-200 stream calls

Migrate LLM Gateway to new Hugging Face Serverless Inference Providers router endpoint

Add URL tracking to LLM connection error log messages

Implement Hugging Face Inference API routing in LLM Gateway and update tests

Add README.md with HF Space metadata

Initial clean deploy