Should we fail and disallow when TSID PK is tried to enable across multiple columns in a table?

[thadguidry]

Personally, I don't see the need to support TSID (time sorted ID aka Snowflake ID) with multiple columns? Almost no point I think.
So maybe we should fail on that... if multiple columns on a table are chosen for TSID enablement?

Because a single TSID PK value can uniquely identify a row... just like a set of 2+ UNIQUE Primary Key columns might be able to also.

TSIDs let you get away with only having a single column as the PK. And guarantee (for non-realtime scientific db's at least) unique row identifiers. (scientific db's typically can want to use nanosecond - not only millisecond)