BUG: invalid dtypes should raise

[jreback]

xref dask/dask#1965

when an explict dtype is provided, we should raise if its not a valid numpy/pandas dtype.

In [1]: pd.Series([], name='time',dtype=pd.Timestamp)
Out[1]: Series([], Name: time, dtype: object)

[petioptrv]

Hi, I am looking into contributing to open source projects and this looks like a good first-time contibution to start with. I'd like to give it a try.

[jreback]

great

contributing docs are http://pandas.pydata.org/pandas-docs/stable/contributing.html

[trevorpburke]

Put in a pull request for this issue. This is my first contribution attempt, so please let me know if I'm way off base here.

[petioptrv]

Should be addressing this at the source of the problem (raise anytime a function passes a non-valid dtype), or should we raise only when this happens in a Series function call?

[petioptrv]

Here is my suggestion for this issue: 607ae9c