COMPAT: alias .to_numpy() for scalars

[jreback]

I think we achieve a bit more consistency in our interfaces if we alias .to_numpy() to .to_datetime64() and .to_timedelta64() for Timedelta and Timestamp respectively.

xref https://github.com/pandas-dev/pandas/pull/24486/files#diff-26a6d2ca7adfca586aabbb1c9dd8bf36L235

[jbrockmendel]

Could do the same for to_py, I think arrow does this

[jreback]

yep

[cbertinato]

What should we do for the to_numpy() keyword arguments? Neither dtype nor copy make sense in the case of a scalar. Just let it raise the TypeError if either is passed, or accept the arguments and do nothing?

[jreback]

i think it’s ok to have them i. the signature for compatibility (default to None); but they will be ignored