fix: tighten reward/grade bounds to [0.01, 0.99] in models, orchestrator, and tests

fix: clamp scores to [0.01, 0.99] in API boundary, UI, inference, and tests

feat: add legal_research domain, multi-domain switching, and README overhaul

test: add exploit tests, integration test, and CI workflow

fix: remove dead code (EXECUTOR, MySQL, bleach), fix PlannerAgent receiver

fix(rewards+orchestrator): additive efficiency, repetition tracking, verify post-answer

fix(server): CORS, sanitization, concurrency, parameter override, mount at /ui

fix: server tests with mock orchestrator injection - 89/89 passing

feat: complete Agentic RAG Gym implementation