string-representation for Timestamp does not show nanoseconds if present

[cancan101]

See:

In [27]: str(pd.Timestamp(200))
Out[27]: '1970-01-01 00:00:00'

In [28]: repr(pd.Timestamp(200))
Out[28]: "Timestamp('1970-01-01 00:00:00.000000200', tz=None)"

whereas microseconds work fine:

In [30]: str(pd.Timestamp(2000))
Out[30]: '1970-01-01 00:00:00.000002'

[cancan101]

One way to fix this would be to implement isoformat in Timestamp. Another would be to implement __str__ and have that method check if nanos !=0.

[cancan101]

    def isoformat(self, sep='T'):
        base = super(_Timestamp, self).isoformat(sep=sep)
        if self.nanosecond == 0:
            return base

        if self.tzinfo is not None:
            base1, base2 = base[:-6], base[-6:]
        else:
            base1, base2 = base, ""

        if self.microsecond != 0:
            base1 += "%.3d" % self.nanosecond
        else:
            base1 += "%.9d" % self.nanosecond

        return base1 + base2

[cancan101]

I intend to fix this (and add some test showing issue) in #5701