fix: tutorials hardcode stored-sample index 9, crash under low-sample profiles

[Jammy2211]

Overview

Results/database/aggregator tutorial scripts across four workspaces hardcode stored-sample index 9 ("the tenth sample"): samples.parameter_lists[9][k] and, in several, log_likelihood_list[9] / log_prior_list[9] / log_posterior_list[9] / weight_list[9]. Any stored run with fewer than 10 samples — e.g. PYAUTO_TEST_MODE fake searches used by local validation sweeps — raises IndexError. Real-search runs pass, so the break is latent in CI but bites every test-mode sweep (found 2026-08-19 while validating PyAutoLabs/PyAutoFit#1504; evidence on PyAutoLabs/PyAutoFit#1486). autogalaxy_workspace's two aggregator guides were previously part-fixed to [0] but kept the now-wrong "tenth sample" prose and redundantly re-print the first sample.

Plan

• Unify the idiom across the family: print the final sample via [-1] with matching "final sample" prose — always valid for any non-empty sample list, and it demonstrates a different sample from the [0] printed just above.
• Apply to every site the sweep found (8 scripts, 4 repos); one PR per repo.
• Validate: autocti's database chain now runs under PYAUTO_TEST_MODE=1 (fewer than 10 stored samples), so a test-mode run crashes before this fix and passes after — a direct regression proof; autofit's cookbooks/result.py is in its smoke list.

Detailed implementation plan

Affected Repositories

• autofit_workspace (primary — the cookbook the others copy)
• autogalaxy_workspace
• autolens_workspace
• autocti_workspace

Sites (from a full-workspace sweep)

• autofit_workspace/scripts/cookbooks/result.py:428 — parameter_lists[9][2]
• autogalaxy_workspace/scripts/guides/results/database/start_here.py:359 — parameter_lists[9][2]
• autogalaxy_workspace/scripts/guides/results/aggregator/samples.py — [0]-code / "tenth sample"-prose mismatch (4 figure-of-merit prints + 1 parameter print)
• autogalaxy_workspace/scripts/guides/results/aggregator/samples_via_aggregator.py — same mismatch
• autolens_workspace/scripts/guides/results/database/start_here.py:373 — parameter_lists[9][2]
• autolens_workspace/scripts/guides/results/aggregator/samples.py:200,219-222 — [9] × 5
• autolens_workspace/scripts/guides/results/aggregator/samples_via_aggregator.py:146,178-181 — [9] × 5
• autocti_workspace/scripts/dataset_1d/advanced/database/examples/samples.py:103,124-127 — [9] × 5
• autocti_workspace/scripts/dataset_1d/results/examples/samples.py:192,211-214 — [9] × 5

Implementation Steps

1. Replace each [9] with [-1] and each "tenth sample" prose mention with "final sample" (inner parameter indexes [2]/[3] unchanged).
2. In the two autogalaxy aggregator guides, change the part-fixed [0] prints to [-1] and correct the prose.
3. Notebook twins left to release-time regeneration per workspace convention.
4. Suggested branch (all repos): feature/tenth-sample-hardcoded-index.

Validation

• autocti: PYAUTO_TEST_MODE=1 run of database/start_here.py → database/examples/samples.py and results chain — crashes at [9] before, must pass after.
• autofit: smoke cookbooks/result.py.
• autolens/autogalaxy: targeted run_smoke.run_script on the touched guides (real_search outputs from 2026-08-19 validation allow fast resume).

Original Prompt

Click to expand starting prompt

Bug in the workspaces: tutorial scripts hardcode stored-sample index 9 and crash under low-sample profiles. autocti_workspace scripts/dataset_1d/advanced/database/examples/samples.py:103 prints samples.parameter_lists[9][2]; autolens_workspace scripts/guides/results/aggregator/samples.py:200 and samples_via_aggregator.py:146 print samples.parameter_lists[9]. Any stored run with fewer than 10 samples (e.g. test-mode fake searches) raises IndexError; real-search runs pass, so this is latent in CI but breaks local/test-mode validation sweeps (found 2026-08-19). Fix: derive the index from len(samples.parameter_lists) or guard it.