fix: omit unfetchable list blobs from column discovery (#4312)

docs/src/js/classes/Table.md

@@ -143,7 +143,8 @@ containing the new version number of the table after altering the columns.
 abstract blobColumns(): Promise<string[]>
 ```
 
-Blob v2 columns, including nested dotted paths.
+Blob v2 columns addressable by row ID, including struct-nested dotted paths.
+Blobs inside lists are omitted because a row ID does not identify an element.
 
 #### Returns
 
@@ -560,6 +561,8 @@ compaction unless stable row ids are enabled. Results keep input order and
 duplicates. Null blobs are `null`. Empty blobs are empty buffers.
 One missing or deleted row ID rejects the entire batch, including on remote
 tables; no partial results are returned.
+Blobs inside lists cannot be fetched by row ID.
+
 Remote servers limit each request to 1024 row IDs and 64 MiB of blob bytes.
 The client splits requests automatically and reads an individual larger
 blob through the Range route. This method still materializes all bytes in

docs/src/js/functions/blob.md

@@ -14,6 +14,8 @@ Declares a `lance.blob.v2` column.
 
 Query results are descriptors, not payload bytes. Use [Table.fetchBlobs](../classes/Table.md#fetchblobs)
 or [Table.fetchBlobFiles](../classes/Table.md#fetchblobfiles) to read bytes.
+Blobs inside lists can be written and queried as descriptors, but cannot
+currently be fetched because a row ID does not identify a list element.
 
 ## Parameters
 

docs/src/python/python.md

@@ -508,6 +508,10 @@ still work. Queries return descriptors. Call
 One missing or deleted row ID rejects the entire blob-read batch; the APIs do
 not return partial results.
 
+Blobs inside lists can be written and queried as descriptors. They are omitted
+from `blob_columns()` because row ids do not identify individual list elements,
+so the fetch APIs cannot read their bytes yet.
+
 ::: lancedb.blob
 
 ::: lancedb._blob.BlobFile