Reject datetime when fraction digits are too large

[nekevss]

We were unwrapping rather than rejecting on the new to_nanosecond method for Fraction.

Context:
to_nanosecond returns an Option<u32>, which represents the nanoseconds value when the digits are < 9, but returns None if the digits exceed 9.

I think this mostly points towards a doc improvement needed in ixdtf that also links to to_truncated_nanosecond to show the difference between the two cases.