feat(tables): add first-class fixed-size-list GPU columns

[ibgreen-openai]

Goals

Represent arbitrary-width fixed-size GPU table rows as first-class, self-describing storage formats without inventing unsupported vertex formats or introducing a competing table owner.

Changes

• Add public FixedSizeList<Format, Size> typing and canonical fixed-size-list<float32,768> support to the central GPU vector-format decoder.
• Preserve GPUData/GPUVector logical row counts while deriving flattened element counts, complete row widths, padded physical strides, and safe byte ranges.
• Keep fixed-size lists out of generic vertex attributes and constants while preserving ordinary GPUTable schemas, record batches, source provenance, append behavior, and explicit packing.
• Bind only the actual row span for padded storage columns, including final rows without trailing padding.
• Document format grammar, schema typing, row/value counts, ownership, storage-only consumption, and chunk-preserving vectors.
• Add focused Node coverage for common embedding widths, invalid formats, overflow, padded rows, table composition, layout publication, and storage binding spans.

Verification

• nvm use (Node 22.22.1).
• yarn install: the configured package registry returns HTTP 403 for required tarballs; verification reuses the existing isolated dependency tree and CI performs the normal immutable install.
• Focused complete @luma.gl/tables Node suite and package TypeScript declarations.
• yarn lint fix.
• yarn build after final source and formatting changes.
• yarn test-node.
• CI=true yarn test with real Chromium/WebGPU coverage.
• yarn website:build.
• (cd website && yarn build).

Risks

Fixed-size lists are intentionally storage-only. Existing one-to-four-component vertex formats, variable-length lists, explicit vertex-layout expansion, ownership, and source batch boundaries remain unchanged.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: cfdbe49b75

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".