label manager: Remove data directory and document how to add label files

Users should now mount label CSV files at runtime using volume mounts
(Docker) or init containers with cloud storage (Kubernetes).

Changes
- Removed COPY data/ line from both Dockerfiles
- The /data directory is still created (mkdir -p /app /data) but empty
- Updated .gitignore to ignore entire data/ directory
- Removed data/** trigger from docker-publish workflow
- Added comprehensive docs/label_manager.md with:
  * Docker volume mount examples
  * Kubernetes init container pattern (recommended for large files)
  * ConfigMap examples (for small files <1MB)
  * PersistentVolume examples (for shared access)
  * Performance considerations and troubleshooting

Author: fordN

.github/workflows/docker-publish.yml

@@ -7,7 +7,6 @@ on:
     paths:
       - 'src/**'
       - 'apps/**'
-      - 'data/**'
       - 'Dockerfile*'
       - 'pyproject.toml'
       - '.github/workflows/docker-publish.yml'

.gitignore

@@ -54,10 +54,8 @@ htmlcov/
 uv.lock
 
 # Data directories (local development)
-data/*.csv
-data/*.parquet
-data/*.db
-data/*.lmdb
+# Large datasets should be downloaded on-demand or mounted via ConfigMaps
+data/
 
 # Build artifacts
 *.tar.gz

Dockerfile

@@ -68,9 +68,11 @@ COPY --from=builder /usr/local/bin/uv /usr/local/bin/uv
 # Copy application code
 COPY --chown=amp:amp src/ ./src/
 COPY --chown=amp:amp apps/ ./apps/
-COPY --chown=amp:amp data/ ./data/
 COPY --chown=amp:amp pyproject.toml README.md ./
 
+# Note: /data directory is created but empty by default
+# Mount data files at runtime using Kubernetes ConfigMaps or volumes
+
 # Install the amp package in the system Python (NOT editable for Docker)
 RUN uv pip install --system --no-cache .
 

Dockerfile.snowflake

@@ -63,9 +63,11 @@ COPY --from=builder /usr/local/bin/uv /usr/local/bin/uv
 # Copy application code
 COPY --chown=amp:amp src/ ./src/
 COPY --chown=amp:amp apps/ ./apps/
-COPY --chown=amp:amp data/ ./data/
 COPY --chown=amp:amp pyproject.toml README.md ./
 
+# Note: /data directory is created but empty by default
+# Mount data files at runtime using Kubernetes ConfigMaps or volumes
+
 # Install the amp package (system install for Docker)
 RUN uv pip install --system --no-cache --no-deps .