Backend array storage: make _data the single source of truth (RasterLayer refactor PR 2)

[Tejasv-Singh]

Backend array storage: _data as the single source of truth

Part of #328. Follows #330 (merged).

This is PR 2 of the GSoC 2026 RasterLayer refactor. Its scope was narrowed following maintainer feedback: Cell remains a mesa.Agent subclass in this PR. Decoupling Cell from Agent has been deferred to a follow-up PR (to land alongside the cell-collection work).

What this does

• Makes RasterLayer._data (a dict[str, np.ndarray]) the single source of truth for all band data.
• Removes the per-cell dual-write loops in set_band, get_band, remove_band, apply_raster, and get_raster, these now read from and write to _data directly instead of scattering/gathering across every cell.
• Makes Cell a thin proxy: __getattr__/__setattr__ route band access (cell.<band>) directly into self._layer._data[name][row, col]. Cells no longer hold per-cell copies of band values.
• Wires cell._layer via a _layer= kwarg passed in _initialize_cells, preserving exact backward compatibility for both the legacy (pos/indices) and current Cell constructor signatures.
• Adds a regression test (test_agent_init_proxy_pos_interaction) confirming that the proxy safely ignores the default attribute assignments made by Agent.__init__, and that pos still behaves.
• Adds a grid_pos property on Cell as a preferred replacement for the deprecated pos (see open question below).

Deferred to follow-up PRs

• Decoupling Cell from mesa.Agent : moved to a separate PR to land with cell collection, so activation isn't split between agents and the collection.
• Examples update : the GIS examples (urban_growth, rainfall, population) still assign band attributes (self.<band> = None) in __init__ before their bands exist. This is handled in a separate mesa-examples PR, per the breaking-change workflow (release → bump requirements.txt → update examples).

CI notes

• Test GIS examples is expected to fail on this branch. It's the deferred examples regression above, not an accidental break, it'll go green once the mesa-examples PR lands after the next release.
• All mesa-geo unit tests pass.

Open questions for review

• grid_pos: kept here for now, but since pos deprecation is conceptually part of the cell redesign, happy to defer it to the decoupling PR if preferred.
• Pre-assigned attribute shadowing: a subclass that assigns self.<band> = None before the band exists will shadow the proxy (the value stays in __dict__). Following the discussion, this is being handled by investigating a migrate-and-erase approach rather than a __getattribute__ safety net; noting it here so it's tracked in review..

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 20b13d18-8e14-4296-9f0a-39fc5a282179

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 78.50%. Comparing base (16ad316) to head (bc00b98).
⚠️ Report is 7 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #332      +/-   ##
==========================================
+ Coverage   78.32%   78.50%   +0.18%     
==========================================
  Files          10       10              
  Lines        1024     1033       +9     
  Branches      168      161       -7     
==========================================
+ Hits          802      811       +9     
  Misses        181      181              
  Partials       41       41              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.