feat: Nemo IPython kernel + JupyterLab devcontainer

[Copilot]

Adds an interactive Nemo Datalog kernel for JupyterLab and a devcontainer that builds the full stack on startup.

nemo-jupyter/ — kernel package

• nemo_kernel/kernel.py — NemoKernel (extends ipykernel.kernelbase.Kernel): accumulates rules across cells, runs the reasoner after each cell, and prints @output predicate results as tables. Parse errors roll back the offending cell. %%reset clears accumulated state.
• nemo_kernel/kernel.json — kernelspec pointing at python -m nemo_kernel
• nemo_kernel/install.py — nemo-kernel-install CLI; registers the spec via jupyter_client with --sys-prefix or --user
• pyproject.toml — minimal package; runtime dep is ipykernel>=6 only (assumes nmo_python is installed separately via maturin)

.devcontainer/

• Dockerfile — rust:1-bookworm base; installs Python + maturin; builds the nmo_python Rust extension wheel from workspace sources; installs JupyterLab, the Nemo kernel, and the default Python ipykernel
• devcontainer.json — forwards port 8888, auto-opens browser, runs jupyter lab on container start, wires VS Code Python/Jupyter extensions to /opt/venv

Example notebook cell (Nemo kernel)

data(1, 2) .
data(hi, 42) .
data(hello, world) .

calculated(?x, !v) :- data(?y, ?x) .

@output calculated .

Results are printed inline after execution. Rules accumulate across cells; use %%reset to start over.