When I use Snowflake to query parquet data directly in S3, it has no problem recognizing my INT96 columns as timestamps. However, when I try to load my S3 data into a local stag
