Summary

• Allow Julia notebook bootstrap in Colab to continue on Julia manifest/runtime version mismatches by default, so hosted Colab runtime updates can re-resolve instead of stopping before setup.
• Update the Python 4-DWave notebook to use dwave.samplers.SimulatedAnnealingSampler instead of the legacy neal import path.
• Add Colab runtime smoke infrastructure: focused Makefile targets, CI steps, and a smoke script for fresh Ocean SDK imports and Julia mismatch behavior across all Julia Colab notebook projects.
• Document the local Colab runtime smoke targets and add unit coverage for the new notebook/bootstrap/test infrastructure expectations.

Tests run

• python -m unittest tests.test_notebook4_pair_sync tests.test_verify_notebooks - passed, 28 tests.
• ~/.local/bin/uv run --group qubo python -c "from dwave.samplers import SimulatedAnnealingSampler; print(SimulatedAnnealingSampler.__name__)" - passed, printed SimulatedAnnealingSampler.
• make verify-julia-colab-mismatch-smoke - passed; real notebook envs 1-MathProg, 2-QUBO, 3-GAMA, 4-DWave, and 5-Benchmarking warned and continued under Julia 1.12.0, then the synthetic strict-mode manifest failed with the expected message.
• make verify-colab-python-runtime-smoke - passed; temporary venv installed only dwave-ocean-sdk and imported/sampled with dwave.samplers.
• ~/.local/bin/uv run --group docs python -m unittest discover -s tests - passed, 62 tests.
• ~/.local/bin/uv run --group docs jupyter book build --html --ci - passed; built 13 pages. The build still reports existing link warnings in several Python notebooks.
• git diff --check - passed.

Notes about tests not run

• I did not run the full notebooks inside hosted Google Colab. The local validation uses Colab detection env vars with a mismatched Julia binary to exercise the failing bootstrap path.
• I did not run the D-Wave quantum annealing credential cells because they require an interactive D-Wave account/token setup.

Closes #9