Fix corner cases of cut with duplicated breaks

[nalimilan]

Apply more systematically the rule that all intervals are closed on the right left and open on the left right except the last one. Throw an error when this would lead to empty intervals unless allowempty=true. Handle 0.0 more systematically.

Fixes #382.

I'm not sure what's the best way of handling -0.0. The current state of the PR treats it as different from 0.0, since that's consistent with our using isless (via searchsortedlast). But we could probably implement a different behavior if we wanted to. 0.0 can be confusing, but if users have it in their data they may wish to preserve it.

I should also check for NaN and throw an error, which is what happened before this PR though it's undocumented and untested.

[bkamins]

Apply more systematically the rule that all intervals are closed on the right and open on the left except the last one.

It is the opposite. Intervals are closed on the left and open on the right (except the last one). But it is OK.

[nalimilan]

Let's check whether Nanosoldier works on two packages:
@nanosoldier runtests(["DataFrames", "MLJ"])

[nanosoldier]

Your job submission was not accepted. Consult the server logs for more details (cc @maleadt).

[maleadt]

Should work now:

@nanosoldier runtests(["DataFrames", "MLJ"])

[nanosoldier]

Update on PkgEvalJob cc547a3 vs. bbac8dc: Accepted

[nanosoldier]

Update on PkgEvalJob cc547a3 vs. bbac8dc: Running

[nanosoldier]

The package evaluation job you requested has completed - no new issues were detected.
The full report is available.