docs: fix dead internal references in script docstrings

[Jammy2211]

Part of PyAutoLabs/autolens_workspace#377, which tracks this sweep across all 7 repos.

Summary

The README half of this drift class is fixed and now gated at PR time (PyAutoHands#213). This is the remaining surface: dead file/folder references in scripts/**/*.py docstrings and comments — the same restructure debt, in prose the gate deliberately does not cover.

Driven by pyauto-brain hygiene refs --json, which reported 129 findings across 7 repos.

Scripts Changed

Docstring/comment prose only — no code, no API, no behaviour. Notebooks and the navigator catalogue are regenerated, since docstring text feeds notebooks/, llms-full.txt and workspace_index.json.

Verified re-points (113):

• log_likelihood_function → likelihood_function — the scripts drop the log_ prefix
• notation/label.yaml → notation.yaml — the config went from a folder to a file
• config/generag.yaml → config/general.yaml — autolens→autogalaxy clone residue, here in script prose
• autofit_workspace/*/plots → plot — the package is singular
• feature/pixelization/… → features/…; subhalo/detection → subhalo/detect
• guides/source_science → the topic packages (source_science.py lives in imaging/, interferometer/, group/ — never guides/)
• preprocess / propocess → imaging/data_preparation
• modeling/imaging/{customize,searches} → guides/modeling/…; imaging/advanced/database → guides/results/database
• simulators → simulator (singular)
• dataset directories quoted without their dataset/ prefix — the folder is written by a simulator, so the fix is qualification, not a restore

Eleven needed prose rewriting, because the target does not exist in any form:

• autogalaxy has no CPU-fast pixelization example → now references the autolens one explicitly
• point_source has no likelihood-function walkthrough → points at cluster/likelihood_function, which is literally titled "Log Likelihood Function: Cluster Point Source"
• no subhalo/detect/examples folder exists anywhere → dead sentence removed
• mat_wrap.yaml was retired in favour of general.yaml
• config/priors has no default subfolder
• z_projects/ is not part of this workspace → the Euclid pipeline has its own repo

Validation

• hygiene refs: 129 → 0 findings across all 7 repos.
• check_navigator.py --banners=fail: PASS on all 6 gated repos.
• Two of my own re-points were wrong and caught by re-running the scanner: I had mapped point_source/log_likelihood_function → point_source/likelihood_function and subhalo/detection/examples → subhalo/detect/examples, but neither target exists. Both re-fixed above.

Known gap

autocti_workspace notebooks are not regenerated. generate.py rejects the autocti project — it is absent from COLAB_PROJECTS and from PyAutoNerves's _PROJECTS registry. Attempting it deletes the notebooks/ tree and aborts (restored here; no damage in this branch). The 5 script fixes stand; that repo's notebooks stay as they were. Filing a follow-up to register autocti.

🤖 Generated with Claude Code