Summary: Issue with timezones not being stored in PG database so changing node timezone had weird effects on HA failovers. By storing the bigint directly we have no ambiguous conversion from pg timestamp to java timestamp and always rely on epoch millisecond time.

Test Plan:
Set up HA with active (A)/standby (B) on UTC

Change standby (B)  to America/Chicago

Promote standby (B)

Change new standby (A) from UTC to America/NewYork

Promote standby (A)

Reviewers: nsingh

Reviewed By: nsingh

Subscribers: sanketh, yugaware

Differential Revision: https://phorge.dev.yugabyte.com/D41534