Update to MADSci v0.8

[RyanTheRobothead]

Summary

Lifts the module from v0.6 to v0.8 for the RAPID 446 grimm rebuild.

• pyproject.toml + pdm.lock: pin madsci.{node_module,client,common}~=0.8.0; bump requires-python to >=3.10.
• src/sealer_rest_node.py: rename self.node_definition.* → self.node_info.* (v0.7 API rename) at the two nest + seal_roll resource-creation sites in create_resources().
• src/sealer_interface.py: rename sealer_plate_deck → nest to match the v0.8 resource-naming convention used elsewhere in the stack.
• notebooks/test_interface.ipynb: small touch-ups aligning with the rename.
• Dockerfile rewritten on the v0.8 pattern:
  • Pin to ghcr.io/ad-sdl/madsci:v0.8.0.
  • uv pip install --python ${MADSCI_VENV}/bin/python -e ./a4s_sealer_module so the editable install lands in the venv the runtime actually uses (system pip lands in /usr/lib site-packages, invisible to the venv interpreter the entrypoint runs).
  • Bake in a dialout_fedora group at GID 18 alongside the container's built-in dialout (GID 20) so the same image works on Ubuntu (dialout=20) and Fedora (dialout=18) for /dev/ttyUSB* passthrough without a per-host group_add: in compose — which gets stripped by the madsci entrypoint's userdel/useradd cycle.
  • Explicit CMD ["python", "a4s_sealer_module/src/sealer_rest_node.py"] (previous default pointed at a non-existent path).

Branched off the existing madsci branch (open PR #5), so this PR's diff includes that whole transition. Reviewers: feel free to close #5 in favor of this, or rebase as you prefer.

Test plan

Verified end-to-end on grimm with sealer_sassy:

• curl /status → ready: true
• curl /state → firmware 515A, current temp 178.3°C, 277 historic sealing cycles
• Container logs show heating block setting accepted (Command *00DH=0175zz! executed successfully) and seal time configured (*00DT=0030zz!)
• ULID 01K9DKWEV7KD1FS72RMEYMEGTZ preserved via .madsci/registry.json