Support partial metadata based aggregation when some projection aggregations are not metadata compatible

[adasari]

Metadata-based aggregation is currently applied only when all projection aggregation functions are metadata-compatible.

If even one aggregation function in the projection is not metadata-based, the query engine falls back to scan-based execution for all aggregations, including those that could have been computed using metadata.

• Metadata-compatible aggregations (e.g., COUNT, MIN, MAX)
• Non-metadata-compatible aggregations (e.g., SUM, DISTINCT, or others)

This results in missed optimization opportunities and unnecessary full segment scans.
Eg:

SELECT COUNT(*), MAX(col1), SUM(col2)
FROM my_table;

This PR adds partial metadata based aggregations for aggregation plan node only.

Pending:

• Unit tests

[adasari]

@Jackie-Jiang Please let me know your thoughts on the changes. I’ll add unit and integration tests once the approach is confirmed. thanks in advance.