fix: restore release result contracts

[Jammy2211]

Summary

Restore release-fidelity result generation, centralize results workflow setup through _quick_fit.py, preserve the fixed tenth-sample tutorial contract, and correctly load saved masked imaging datasets.

Scripts Changed

• scripts/guides/results/_quick_fit.py — now creates two capped Nautilus fits in plain output/results_folder, retains all 300 samples, writes latent summaries, and disables fast-plot skipping so PNG/FITS workflow products exist.
• scripts/guides/results/workflow/{csv_make,png_make,fits_make}.py — reuse _quick_fit.py instead of duplicating their own two-fit setup; aggregate from results_folder instead of results_folder_csv_png_fits.
• scripts/guides/results/aggregator/* — use plain Path("output") / "results_folder"; no workspace-side test-mode path handling.
• scripts/guides/results/aggregator/galaxies_fits.py — correct prose to match tenth-last sample access.
• scripts/guides/results/start_here.py — use results_folder for the quick workflow output path and allow masked zero-valued noise pixels when reloading saved dataset FITS.
• config/build/env_vars_release.yaml — unset PYAUTO_TEST_MODE for results guides so they run outside PyAuto test mode.
• Corresponding notebooks and generated catalogues regenerated from scripts.

Test Plan

• Four target release-profile scripts pass independently
• csv_make.py creates all seven expected CSV files
• Clean release-profile _quick_fit.py passes with PYAUTO_TEST_MODE unset, 300 retained rows, and no output/test_mode/results_folder
• Clean release-profile workflow run passes: csv_make.py, png_make.py, fits_make.py reuse two centralized fits with latent summaries, fit.png, and fit.fits under output/results_folder; no results_folder_csv_png_fits is created
• Focused workflow rerun confirms no with_test_mode_segment, PYAUTO_TEST_MODE, or output/test_mode references remain in scripts/guides/results
• Workspace smoke tests: 9/9 passed

🤖 Generated with Codex

[Copilot]

Pull request overview

This pull request restores the expected “release-fidelity” results outputs for the results tutorials, ensuring the tutorials can reliably demonstrate indexed sample access and reload saved FITS datasets without failing noise-map validation.

Changes:

• Disable noise-map validation when reloading saved dataset.fits products in the results start_here guide (check_noise_map=False) to allow masked zero-valued noise pixels.
• Ensure the results quick-fit helper retains all samples by disabling output-side weight threshold filtering (samples_weight_threshold = None).
• Align results tutorial prose with the existing “tenth-last sample” indexing contract and adjust release env vars to run results guides with PYAUTO_TEST_MODE=0.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
scripts/guides/results/start_here.py	Loads saved FITS dataset products with check_noise_map=False to support masked zero-noise pixels.
scripts/guides/results/aggregator/galaxies_fits.py	Updates narrative text to match tenth-last sample indexing used in code.
scripts/guides/results/_quick_fit.py	Disables sample-weight threshold filtering so results guides can demonstrate indexed sample access over the full capped sample set.
notebooks/guides/results/start_here.ipynb	Regenerated notebook reflecting the check_noise_map=False change from the script.
notebooks/guides/results/aggregator/galaxies_fits.ipynb	Regenerated notebook reflecting the prose update from the script.
notebooks/guides/results/_quick_fit.ipynb	Regenerated notebook reflecting the config override to retain all samples.
config/build/env_vars_release.yaml	Forces PYAUTO_TEST_MODE=0 for results guides in the release profile to preserve expected sample/output contracts.