[codex] Improve effect handling for mesh-backed layers

[charlieforward9]

Adding extra attributes has a cost and will slow down the layer even if they are unus…#9933 (comment)

Direction

This is a draft exploration of deck.gl effect handling for globe-facing rendered surfaces.

The current patch started with BitmapLayer because it exposed the issue quickly in GlobeView: lighting and material effects need layer implementations to provide the right geometry-space inputs, normals, and shader hooks instead of assuming every renderable surface behaves like existing mesh-backed layers.

That said, I do not want to pursue a BitmapLayer/TileLayer-centered solution as the merge path. BitmapLayer is fundamentally a 2D image layer, and fixing globe lighting by special-casing bitmap tiles risks spending review energy on the least useful target.

Desired Merge Path

The useful upstream direction is to improve effect handling for layers that are already mesh/model-oriented, especially:

• TerrainLayer
• mesh-backed composite layers
• model/scenegraph-style layers
• other layers where lighting/material support is a natural part of the rendering model

The goal is to make lighting and effect behavior more predictable for globe and terrain use cases without coupling the fix to TileLayer or treating bitmap imagery as the primary lighting surface.

Current State

This draft currently contains an opt-in BitmapLayer lighting prototype:

• adds BitmapLayer material prop support
• wires the Phong lighting shader path into BitmapLayer when material is enabled
• keeps existing BitmapLayer behavior by defaulting material to false
• updates BitmapLayer docs/tests/goldens around that prototype

Treat that as a reference point for the effect plumbing problem, not as the final scope I want to merge.

Not Pursuing

I am closing the separate TileLayer/GlobeView bitmap-coordinate follow-up and do not want to carry that as a second PR. The TileLayer-specific 2D imagery path should not become the center of this work.

Validation So Far

• BitmapLayer-focused render/test coverage was added with the prototype.
• The separate TileLayer follow-up was validated locally, but it is being closed rather than pursued.

Progress

Keep this PR as a draft while the scope is redirected toward TerrainLayer and other mesh/model-related layers.

visgl/deck.gl#10251
visgl/deck.gl#10306
visgl/deck.gl#10350

[coveralls]

coverage: 83.401% (+0.01%) from 83.39% — codex/bitmap-layer-lighting into master

[charlieforward9]

@copilot better to preserve existing behavior. Modify in v10?

Suggested change

	* @default true
	* @default false