add content moderation task with real OpenAI Moderation data

update README with alignment task details and issue breakdown

add alignment data QA task: 12 issues in LLM instruction-tuning data

Fix port to 8000 for validator compatibility

remove base_path: /web to fix HF Space iframe 404

remove binary PNGs for HF push compatibility

use port 7860 for HF Spaces compatibility

minor change for meeting requirement format

expand datasets to include harder real-world scenarios

add fix stage+demo

fixes v1: add per step reward

init