gh-107773: Make datetime subclass repr consistent both implementations

[donBarbos]

New behavior:

>>> class A(_datetime.time): pass
...
>>> class B(_pydatetime.time): pass
...
>>> repr(A(1))
'A(1, 0)'
>>> repr(B(1)) # or str()
'B(1, 0)'

Old behavior:

>>> class A(_datetime.time): pass
...
>>> class B(_pydatetime.time): pass
...
>>> repr(A(1))
'A(1, 0)'
>>> repr(B(1)) # or str()
'__main__.B(1, 0)'

• Issue: datetime subclass repr varies by implementation #107773

[mattip]

Might be nice to add the reproducer from #107773 as a test, if only to make sure this does not regress and that the C and Python implementations remain in sync.