You misinterpret my comment as being against databases, when in fact I'm simply against the notion of utilizing a purpose-built OLAP database (Clickhouse) merel

You misinterpret my comment as being against databases, when in fact I'm simply against the notion of utilizing a purpose-built OLAP database (Clickhouse) merely as the storage layer for yet another analytical platform (Pandas). In this hypothetical scenario, before the data ever makes it into Clickhouse, there's a very high likelihood it would already live in an OLTP database such as Postgres or perhaps parquet files living on S3 designed to be read by Athena or Redshift Spectrum etc.