- Replace the 'Anisotropy / GPM Support' row in the comparison table:
  DisSModel provides neighborhood definition via libpysal spatial weights
  (Queen, Rook, KNN, custom adjacency), not a GPM implementation
- Remove the runtime 'TOML-based model registry' claim from the executor
  description (platform-side feature); keep what the package provides:
  executors as standard Python packages with parameters resolved from a
  version-controllable TOML specification
- Clarify that Streamlit dashboards live in the satellite packages; the
  core ships Streamlit-compatible input widgets
- Mark the DisSModel Platform explicitly as a separate project in the
  Research Impact section

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

…chmark results

BR-MANGUE now lives in the brmangue-dissmodel package, which holds both
the raster and vector substrates plus two validation levels (raster vs
TerraME and vector vs raster). coastal-dynamics is being retired.

- paper: cite brmangue-dissmodel (@BRMangue) instead of @CoastalDynamics;
  describe the two-level validation — raster vs TerraME on the Maranhão
  Island dataset (50,496 cells, 20 steps: uso 99.90%, solo 99.85%, alt
  90.67% within 1 mm) and vector vs raster on the 60x60 synthetic grid
  (10 steps, 100% match on all bands, raster 5.0 ms/step vs vector
  124.4 ms/step)
- paper: refresh Game of Life benchmark table with current measurements
  (1M cells at ~27 ms/step)
- paper.bib: replace the CoastalDynamics entry with BRMangue
- README and docs/examples.md: point the ecosystem tables to
  brmangue-dissmodel (install verified via pip git+ and smoke imports)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

JOSS requires papers between 750 and 1750 words; after the BR-MANGUE
two-level validation was added the text reached 1795 words. Three
surgical cuts bring it to 1745 without losing citations or required
content (non-specialist explanation and research applications kept):

- drop the redundant 'hybrid data types' sentence in Statement of Need
  (dual-substrate architecture is already described in the Summary)
- compress the TerraME/LUCCME counterpart sentence in Software Design
- tighten the ExperimentRecord provenance sentence in Performance

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Doc/readme platform scope