feat: Phase 5 — GitHub Actions for lint + profile re-run

[Jammy2211]

Summary

Phase 5 (final) of the autolens_profiling z_feature roadmap — wires up CI for the new repo and threads AUTOLENS_PROFILING_SMOKE=1 through every profile script so the lint workflow's smoke step is cheap.

Closes #7.

What lands in this PR

Two GitHub Actions workflows

File	Trigger	Job
.github/workflows/lint.yml	pull_request, push: main	ruff lint + format, dashboard idempotence (build_readme.py --check), lychee link-rot, smoke 3 scripts
.github/workflows/profile.yml	workflow_dispatch (with sections filter), release: published	Run profile scripts, regenerate results/, refresh dashboard, commit back as github-actions[bot] with [skip ci]

Both target ubuntu-latest (CPU-only). The smoke step short-circuits each profile script via AUTOLENS_PROFILING_SMOKE=1, so no real artifacts are produced in the lint workflow.

ruff.toml at repo root

This repo isn't a Python package, so a standalone ruff.toml (a supported configuration location) is cleaner than fabricating a pyproject.toml. Conservative rule selection: E/F/W/I/UP/B; E501/E402/F401/B008 ignored for scientific code patterns.

AUTOLENS_PROFILING_SMOKE=1 short-circuit threaded into 17 scripts

Inserted via an AST-driven helper at the first non-import, non-docstring top-level statement of each script:

import os as _smoke_os
import sys as _smoke_sys
if _smoke_os.environ.get(\"AUTOLENS_PROFILING_SMOKE\") == \"1\":
    print(f\"[smoke] {__file__}: imports + module setup OK; exiting.\")
    _smoke_sys.exit(0)

Coverage: 17 scripts across likelihood/{imaging,interferometer,point_source,datacube}/*.py, simulators/*.py, searches/nautilus/*.py. Idempotent — re-running the helper detects existing instances.

Workflows README

.github/workflows/README.md documents both workflows, the manual-trigger procedure, and captures all design decisions inline for future maintainers.

Design decisions resolved

• Runners: CPU-only on github-hosted ubuntu-latest. Self-hosted GPU runners deferred (additive on this workflow shape — see top-level README "Future enhancements" for the matching dashboard-side column extension).
• Cadence: workflow_dispatch + release: published. No weekly cron — releases are the natural cross-version-comparison anchor.
• Bot identity: github-actions[bot] with [skip ci] subject line. Avoids lint workflow re-trigger on the auto-generated commit.
• Failure handling: continue-on-error: true per section. A single regression emits ::warning:: and leaves the dashboard cell as ERR rather than blocking the refresh for the other 16+ scripts.
• Smoke env var: AUTOLENS_PROFILING_SMOKE=1, per the Phase 5 prompt.
• ruff config location: ruff.toml at root (no pyproject.toml because this isn't a Python package).
• Skipping simulators/point_source.py in the profile loop: its default dataset_name=\"simple\" overwrites Phase 1's tracked likelihood JSONs. Manual runs with a non-conflicting dataset_name remain supported.

Test plan

• yaml.safe_load PASSES on both .github/workflows/*.yml.
• py_compile PASSES on all 17 SMOKE-instrumented scripts.
• AUTOLENS_PROFILING_SMOKE=1 python likelihood/imaging/mge.py exits 0 in <1s with the expected smoke-OK line.
• Same for simulators/imaging.py and searches/nautilus/simple.py.
• python scripts/build_readme.py --check exits 0 after SMOKE insertions — Phase 4 idempotence is preserved.
• First real CI run will be against this PR itself — local YAML parsing is necessary but not sufficient. Any GitHub Actions runtime issues get follow-up PRs.

Refs

• z_feature parent: feat: Phase 1 — mirror JIT likelihood profiling scripts #1.
• Phase 1 (shipped): feat: Phase 1 — mirror JIT likelihood profiling scripts #2; F1 re-mirror fix: re-mirror likelihood scripts and datasets from clean origin/main of _developer #3.
• Phase 2 (shipped): feat: Phase 2 — mirror simulator profiling scripts #9.
• Phase 3 (shipped): feat: Phase 3 — mirror Nautilus profiling scripts #8.
• Phase 4 (shipped): feat: Phase 4 — instrument-framed README dashboard #10 — provides scripts/build_readme.py which both workflows call.
• F1 upstream issues: PyAutoLens#514, _developer#68, _developer#69 — profile.yml's continue-on-error ensures their failure surfaces as ERR cells rather than blocking the dashboard refresh.

🤖 Generated with Claude Code