fix: HF Space port + TRL install order in training Dockerfile

fix: correct model name to unsloth/gpt-oss-20b (no -instruct suffix)

fix: move unsloth import first, make wandb optional via WANDB_API_KEY

fix: install trl>=0.16 last with --upgrade to beat unsloth dep pins

fix: use CUDA devel image and pin vLLM to 0.12.0

fix: strip ANSI codes in _run_tests() so ✓/✗ count correctly

feat: add W&B reward logging to both training scripts

feat: RFC 005 interactive rollout wrapper + multi-turn GRPO training

feat: replace handcrafted user_messages with real MOA session traces

feat: multi-turn tool-using GRPO training

feat: multi-turn tool-using RL environment (RFC 005 pattern)

fix: add make+g++ for node-pty native build

feat: use real moav2 source as RL env, symlinked sandbox, demo.py

feat: use real moav2 source as RL task suite — symlinked sandbox, 3 real service tasks

fix: embed task content directly, self-contained vitest sandbox

upgrade: gpt-oss-20b BF16, vLLM GRPO, Northflank env URL, max_steps=300

add training/ as real directory (Dockerfile + train.py)

add GRPO training job (Llama 3.1 8B + Unsloth + TRL)

fix: add README with HF Space metadata

add .gitignore, remove build artifacts

initial moa rl environment