The Stack Exchange Data Explorer has a lot of useful data in its Votes table. Unfortunately, a lot of this data, and particularly the timing data, can be used to deanonimize voting patterns, which is definitely unwanted, so to this end vote timing information is coarse-grained to 1-day intervals. For upvotes and downvotes this is somewhere between a Good Thing and an essential for the site's functioning.
However, there's a bunch of other vote types in that table,
2. UpMod (AKA upvote)
3. DownMod (AKA downvote)
5. Favorite (UserId will also be populated)
6. Close (since 2013-06-25 close votes are ONLY stored in the PostHistory table)
8. BountyStart (UserId and BountyAmount will also be populated)
9. BountyClose (BountyAmount will also be populated)
and for a lot of those (a) timing information is public anyway (Bounty Start, Bounty Close) or carries no deanonimization risk (Accepted, Delete, Undelete). This timing data won't save the world by itself, but it is relatively useful for relatively interesting queries one can think of. (As an example, answers posted within n hours of a bounty being closed / awarded, or time-to-accept statistics over all questions.)
Is there some specific reason why all timing data is pared down, other than uniformity? Would it be very hard to make these sorts of vote types have precise timing data available in the data dump and data explorer?