[remsky]

This PR replaces the placeholder ML training example with one that runs.

Covers both items in 138: epoch-based training and a human review.

Closes #138

Changes

examples/ml-training/application.py trains a scikit-learn model:

• One epoch is one action, with a checkpoint for each epoch.
• The stop rule is a transition condition.
• The run halts before human_review and takes the decision as an input.
• With continue, the reviewer can pass a new learning_rate.
• A SQLite persister lets a new process resume the run by app_id.

Around it:

• The required example files (notebook, README, diagram, requirements.txt) and a test.
• ml-training is no longer in the skip list in validate_examples.py.
• The docs page describes the new example and drops the WIP notice.
• CI: one step in validate-examples installs the requirements of the example and runs its test.

How I tested this

• Clean install of Burr 0.42.0 from requirements.txt: the test, a terminal run with stop and resume, notebook all passed.
• Source tree: pre-commit (without frontend-lint-staged, no UI file changed) and validate_examples.py passed.
• Windows with Python 3.12 only.

Notes

• I used scikit-learn so that the example is lightweight. Let me know if you'd rather I swap to PyTorch.
• The CI step is there because .[tests] does not include scikit-learn. I can remove it if you prefer.
• docs: fix dead Lightning callbacks and cortex links #1009 edits the README and the docs page of this example; if it merges first it should only be a small mechanical conflict.

Follow-up if you'd want it: would be pretty clean to drop in an agent action that reads the summary and picks the next learning rate. The human-in-the-loop would then update the agent's instructions or approve the final model.

Checklist

• PR has an informative and human-readable title (this will be pulled into the release notes)
• Changes are limited to a single goal (no scope creep)
• Code passed the pre-commit check & code is left cleaner/nicer than when first encountered.
• Any change in functionality is tested
• New functions are documented (with a description, list of inputs, and expected output)
• Placeholder code is flagged / future TODOs are captured in comments
• Project documentation has been updated if adding/changing functionality.