feat: rework docker workflow for gpu-first training

- rewrite docker.sh with gpu default and --no-gpu override
- inject host uid/gid via ASTRAI_UID/GID in train.sh compose()
- filter readonly UID/GID lines when sourcing .env.train
- build image user via USER_UID/USER_GID args matching host uid/gid
- pass all GPUs (count: all) and filter by CUDA_VISIBLE_DEVICES inside the container
- forward NCCL vars through compose environment

Dockerfile

@@ -57,8 +57,13 @@ COPY docs/ ./docs/
 COPY pyproject.toml .
 COPY README.md .
 
-# Create non-root user
-RUN useradd -m astrai && chown -R astrai:astrai /app
+# Create non-root user matching the host uid/gid (passed via build args)
+ARG USER_UID=1000
+ARG USER_GID=1000
+RUN groupadd -g "${USER_GID}" astrai \
+    && useradd -m -u "${USER_UID}" -g astrai astrai \
+    && chown -R astrai:astrai /app
+ENV HOME=/home/astrai
 USER astrai
 
 ENV PYTHONUNBUFFERED=1 \