More consistent handling of POSIXlt input to setDT

[MichaelChirico]

Closes #4800.

Still needed:

• Figure out what to do about the inconsistency of setDT() for list vs. data.frame input with matrices: setDT should check if any column is POSIXlt #4800 (comment)
• Split out the length_with_dispatch() part which is a vestige of an earlier attempt to get setDT() working on POSIXlt input, but which also happens to be useful (and I think solves?) [BUG] Cannot print a data.table with column of type rvar from posterior package #5981
• NEWS, tests

[github-actions]

Generated via commit ee9f287

Download link for the artifact containing the test results: ↓ atime-results.zip

Task	Duration
R setup and installing dependencies	4 minutes and 43 seconds
Installing different package versions	7 minutes and 44 seconds
Running and plotting the test cases	2 minutes and 21 seconds