Add shared polygon atlas API for strict R/Python ggseg visual parity

[mqqq333]

Description

This PR adds a minimal polygon atlas interoperability API for rendering R ggseg-exported 2D polygon atlas tables in python-ggseg. The goal is to support strict visual parity between R ggseg and python-ggseg cortical gray-matter plots by allowing both backends to render the same atlas geometry and, when desired, the same identity colors via fill_hex.

This PR adds:

• read_polygon_atlas(path) for loading canonical polygon atlas CSV files;
• validate_polygon_atlas(atlas) for schema validation;
• join_polygon_values(...) for explicit join-by-label / join-by-region behavior;
• plot_polygons(...) for rendering shared polygon atlas tables;
• fill_column="fill_hex" identity-color rendering for exact cross-backend color parity;
• tests for schema validation, value joining, unmatched labels, and identity-color plotting;
• documentation describing the R/Python visual parity workflow.

The implementation is additive and does not change existing plot_dk, plot_aseg, or plot_jhu behavior.

Related Issues

No related issue yet.

Type of Change

• Bug fix
• New feature
• Documentation update
• Other (describe): Cross-language visual-parity / interoperability API

Checklist

• I have read the contributing guidelines
• I have run devtools::check() with no errors
• I have added tests for my changes (if applicable)
• I have updated documentation (if applicable)
• My code follows the tidyverse style guide

[mqqq333]

I also tested this locally with a real DK polygon table exported from R ggseg.

The comparison image below uses:

• the same polygon CSV as canonical geometry;
• the same fill_hex column as identity colors;
• Python rendering through the proposed plot_polygons(..., fill_column="fill_hex");
• R rendering from the same polygon table.

Local smoke test using the proposed API:

read_polygon_atlas: OK
rows = 3233
labels = 70
features = 204
unique_fill_hex = 63
rendered_patches = 204

This is why the PR focuses on shared polygon geometry and optional identity-color rendering via fill_hex: exact R/Python visual parity is reliable when
both backends consume the same geometry and colors, rather than reconstructing atlas geometry independently.