[python] Support MAP<K, BLOB> in multimodal BLOB reads

[XiaoHongbo-Hope]

Purpose

MultimodalTable.scan().read_blobs() and stream_blobs() recognize only scalar BLOB fields, so MAP<K, BLOB> columns are rejected as non-BLOB columns.

This change recognizes MAP BLOB fields, flattens all map-value descriptors into the same coalesced range read used for scalar BLOBs, and reconstructs row-aligned key-value pairs afterward. Map key order, null maps, empty maps, null values, inline values, and mixed scalar/MAP reads are preserved.

The multimodal documentation and end-to-end tests cover both materialized and streaming MAP BLOB reads.

Tests

• pytest pypaimon/tests/multimodal_table_test.py -q (68 passed)
• flake8 for changed Python files
• python3.6 -m py_compile for changed Python files
• git diff --check

[JingsongLi]

[P2] Avoid per-cell target bookkeeping for scalar BLOB reads

This sends every non-null scalar BLOB cell through the generic target mapping, retaining two target tuples plus the bodies and ranges lists until all reads complete. In an isolated 300k-descriptor run, peak allocations increased from 41.0 MiB on the base to 94.0 MiB on this head; for inline cells they increased from 26.2 MiB to 70.1 MiB, even though no MAP column was involved. Since read_blobs() can materialize large datasets, this O(rows) metadata regression can add substantial GC pressure or trigger OOMs for existing scalar-only callers. Could we keep the previous contiguous offset/slice path for scalar columns and reserve target mappings for MAP entries, for example by passing the MAP column set into this helper?

[JingsongLi]

+1