Strengthen queue benchmark and refresh landing page

Add cluster-aware queue dynamics to helpdesk env

Upgrade helpdesk env with queue dynamics and operational actions

Add queue-planning helpdesk routing mechanics

Strengthen hard-task investigation and grading

Clamp reported task scores to open interval and match sample logs

Normalize remaining score fields into open interval

Keep invalid-action task scores inside open interval

Clamp grader task scores to open interval

Run all tasks by default and keep task scores inside open interval

Use AWS public ECR mirror for python base image

Pin python base image digest for stable Docker builds

Use evaluator API_KEY for LLM proxy and strengthen env

Add policy learning loop and strengthen RL-style environment

Ignore action metadata in extra field validation

Finalize gap fixes and lightweight competitive upgrades

fix: resolve full test suite failures from PR review

feat: competitive upgrade for hackathon submission

fix port 8000->7860 in app.py/openenv.yaml, add pyproject script entry, fix stubs

fix: expand inference docstring and add git to Dockerfile

Fix dataset loader for UTF-8 BOM on Windows

April 2nd tasks

March 30 - April 1st : sever/

Initial commit