feat: cluster-scale visualization — per-plane critical curves/caustics aplt helpers

[Jammy2211]

Summary

Adds cluster-scale visualization to the library: a new autolens/cluster/plot/ package of module-level aplt helpers (the autolens/weak/plot/ pattern), promoted from the raw-matplotlib prototype in autolens_workspace_test/scripts/cluster/visualization.py. The headline capability is per-source-plane critical curves and caustics for multi-plane tracers: at cluster scale every source redshift has its own curve set (D_LS/D_S differs per plane), which the galaxy-scale single-plane defaults cannot express.

API Changes

Added five aplt-level plotting helpers (plot_positions_overlay, plot_image_group_zooms, plot_critical_curves, plot_caustics, subplot_cluster_dataset) plus the WONG_PALETTE / CLUSTER_CMAP constants in autolens.cluster.plot.cluster_plots. Purely additive — no existing symbol changes.
See full details below.

Test Plan

• test_autolens/cluster/plot/test_cluster_plots.py — 6 tests: one per helper (plot_patch save assertions) plus a physics test asserting the z=2 plane's tangential critical curve is larger than the z=1 plane's (the reason per-plane curves exist).
• Full PyAutoLens suite (count in issue).
• Companion integration script run end-to-end (autolens_workspace_test PR).

Validation checklist (--auto run — in-session directives)

• Effective level: supervised (header: supervised, cap: feature → supervised); one-shot ship authorized in-session by the user, Heart YELLOW acked in-session
• Plan: on the issue (feat: cluster-scale visualization — per-plane critical curves/caustics, large-FoV aplt helpers #577)
• Human: helper API shape right for the flagship Lenstool example to consume?
• Human: merge, amend, or reject — then log the outcome

Full API Changes (for automation & release notes)

Added

• aplt.plot_positions_overlay(positions_list, image=, pixel_scales=, centres=, halo_centres=, redshift=, cosmology=, kpc_scale_bar=, ax=, ...) — full-field per-source-coloured multiple-image positions with conventional cluster markers and a physical scale bar.
• aplt.plot_image_group_zooms(positions_list, image, zoom_arcsec=, max_cols=, ...) — one zoom panel per observed multiple image, colour-framed by source.
• aplt.plot_critical_curves(tracer, grid, image=, plane_indices=, include_radial=, ax=, ...) — tangential (+ radial) critical curves of every source plane via the multi-plane LensCalc (use_multi_plane=True, plane_j=j); numpy path only (contour solvers are not vmap-safe), grid-resolution guidance in the docstring.
• aplt.plot_caustics(tracer, grid, plane_indices=, include_radial=, ax=, ...) — the per-plane caustics in source-plane coordinates (radial on by default — it bounds the central demagnified-image region).
• aplt.subplot_cluster_dataset(positions_list, image=, tracer=, grid=, ...) — positions overlay | per-plane critical curves mosaic.
• autolens.cluster.plot.cluster_plots.WONG_PALETTE, CLUSTER_CMAP — the shared per-source palette / colormap constants.

Migration

• None required (additive). The raw-matplotlib patterns in the old workspace_test prototype are superseded by these helpers.

🤖 Generated with Claude Code