unit of least precision for RR and RDF

[robertwb]

I wrote this while attacking #4746. Didn't turn out to be relevant to that, but it's useful to have on its own anyways.

Component: basic arithmetic

Issue created by migration from https://trac.sagemath.org/ticket/5071

[roed314]

comment:1

Looks good. The only change I would suggest is that the docstring should include the behavior on inf and NaN.

[robertwb]

Attachment: 5071-ulp.patch.gz

[robertwb]

comment:2

OK, doctests updated.

[sagetrac-cwitty]

comment:3

Code looks good, doctests pass.

Excellent idea!

Positive review.