Add column locations within file to DuckLake, aka index Parquet metadata.

[jeschkies]

👋 would it be possible to add column locations to the column table or another appropiate table?

Currently, readers would have to read the Parquet metadata first in order to find the locations for column chunks. This requires to either download the whole file first or issue multiple requests to the bucket storage (please correct me if I missed an alternative).

If the column (chunk) location was indexed it could reduce requests to the bucket storage.

[hpvd]

this is maybe also named in the 2nd link in this comment of general index discussion #389 (comment)