Add AppView support for new array format types and ndarray v1.1.0 annotations

[maxine-at-forecast]

Context

The lexicon repo (forecast-bio/atdata-lexicon#21) has added new array format tokens and shim schemas:

New array format tokens

• sparseBytes — Scipy sparse matrix format (CSR/CSC/COO)
• structuredBytes — Numpy structured array format (compound dtypes)
• arrowTensor — Arrow tensor IPC format
• safetensors — HuggingFace safetensors format

New shim schemas

• schemas/sparse_shim.json — sparse matrix bytes
• schemas/structured_shim.json — structured array bytes
• schemas/arrow_tensor_shim.json — Arrow tensor bytes
• schemas/safetensors_shim.json — safetensors bytes
• schemas/dataframe_shim.json — Parquet-encoded tabular data

NDArray shim v1.1.0

• schemas/ndarray_shim_v1.1.0.json — adds optional dtype, shape, and dimensionNames annotation fields

Tasks

• Update AppView schema validation to accept new arrayFormat known values
• Handle new shim types in schema resolution (resolveSchema XRPC)
• Support ndarray v1.1.0 annotation fields in schema validation/display
• Update any UI components that display array format information
• Add tests for AppView handling of new format types

References

• Lexicon PR: Add multi-format array types with shim schemas and ndarray v1.1.0 atdata-lexicon#21
• Lexicon issue: Add multi-format schema and array type lexicon definitions atdata-lexicon#11