Feature: More of an anti-feature: TIMESTAMP-TZ APIs should be deprecated

[dtenny]

API

C

Description

While building lisp API support based on the C API, I noticed that the semantics of kuzu_prepared_statement_bind_timestamp_tz and kuzu_prepared_statement_bind_timestamp appear to be identical. The comments on their respective structs https://github.com/kuzudb/kuzu/blob/master/src/include/c_api/kuzu.h#L246-L258 indicate that they both accept microsecond deltas from the epoch. Perhaps the tz variant was originally meant to do more, but it seems to interpret the microsecond value as UTC based just like vanilla 'timestamp' interfaces do.

In a discussion with Guodong Jin in Discord he indicated that he spoke with the team that it is probably reasonable to deprecate the timestamp_tz materials. If it is somehow doing more with timezones, then more documentation should be added to describe how the timezone context is provided and used, but that doesn't seem to be the case at this time.