end_of_day/1 should be comparable with inspect

[josevalim]

One way to look at it is we should be able to write the assertion under question using just sigils:

assert NaiveDateTime.end_of_day(~N[2000-01-01 23:00:07]) == ~N[2000-01-01 23:59:59]

and the fact we can't is a reason to revisit this?

Originally posted by @wojtekmach in #12443 (comment)

[josevalim]

We should probably truncate it based on the precision: {(10 ** precision) - 1, precision}

[thiamsantos]

That makes sense. I will open a PR.