5.0

[icrdr]

Summary

This PR updates the volumetric import flow so the expensive import stage runs in a background Blender CLI worker instead of blocking the main Blender UI.

Key changes:

• Register a Blender CLI command, bioxelnodes_import_worker, via bpy.utils.register_cli_command
• Replace direct --python io_worker.py execution with the official CLI command path
• Remove manual sys.path modification from the worker
• Support Blender Extension module names such as bl_ext.xxx.bioxelnodes
• Keep metadata reading synchronous on the main thread to avoid unsupported threading / queue usage
• Run the heavy import stage in a background Blender process:
  • parse volumetric data
  • resize/process layers
  • write layer cache files
  • write progress/result JSON
• Split cache writing into save_layers_to_cache() so the worker can write cache metadata and the main process can merge it back into Blender state
• Improve worker progress messages for startup, module loading, parsing, and processing stages

Design Notes

The final design avoids unsupported patterns:

• no sys.path mutation
• no threading
• no queue
• no direct execution of worker files as standalone scripts

Metadata loading is done synchronously because it is a lower-frequency, lighter step compared with the full import. The expensive import work still runs outside the main Blender process.

Tradeoffs

• The first heavy import still needs to start a background Blender process
• Metadata reading may briefly block the UI
• No persistent worker/server is introduced yet, keeping lifecycle complexity low
• Repeated imports will still start a new background process each time

Verification

uv run python -m compileall src\bioxelnodes\operators\io.py src\bioxelnodes\operators\io_worker.py
git diff --check -- src\bioxelnodes\operators\io.py src\bioxelnodes\operators\io_worker.py