Clean up image size tracking and buffers

[mkeeter]

• Add dedicated DepthImageBuffer to avoid tracking depth separately (it still only contains width * height items)
• Removed size argument from many buffer construction functions, because we always resize them when they're used

[Copilot]

🟡 Changes recommended

There is a confirmed state bug in submit_merge (stale image_count on empty input) plus a misleading depth-omitting size mismatch error message that should be corrected.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

This PR simplifies GPU image-buffer sizing in the WGPU voxel pipeline by introducing a depth-aware buffer type (DepthImageBuffer) and removing explicit size parameters from several effect buffer constructors, relying on automatic resizing at submission time.

Changes:

• Added DepthImageBuffer (a GenericFlexBuffer sized by VoxelSize) to preserve depth metadata without separately tracking it.
• Updated voxel/effects buffer APIs (merge_buffers, shade_buffers, ssao_buffers) to be constructed without an explicit size and resized on first use.
• Generalized Gpu::read_buffer_for to accept any GenericFlexBuffer whose size can convert into ImageSize.

File summaries

File	Description
fidget-wgpu/src/voxel/mod.rs	Switches voxel geometry storage to DepthImageBuffer and updates related sizing/growth logic and tests.
fidget-wgpu/src/voxel/effects/mod.rs	Converts merge/shade outputs to depth-aware buffers and removes size args from buffer constructors.
fidget-wgpu/src/lib.rs	Generalizes read_buffer_for to work with depth-aware buffers via Into<ImageSize>.
fidget-wgpu/src/buf.rs	Introduces DepthImageBuffer alias and implements BufferItemCount for VoxelSize.
demos/cli/src/main.rs	Updates CLI usage to new no-size buffer constructors.
CHANGELOG.md	Documents the API changes and the new DepthImageBuffer type.

Review details

• Files reviewed: 6/6 changed files
• Comments generated: 3
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.