API: add astype coerce argument / remove raise_on_error

[jreback]

http://stackoverflow.com/questions/25952790/convert-pandas-series-from-dtype-object-to-float-and-errors-to-nans/25952844#25952844

maybe removeraise_on_error and rename to coerce (more inline with the rest of pandas API)

In [30]: pd.Series([1,2,3,4,'.']).convert_objects(convert_numeric=True)
Out[30]: 
0     1
1     2
2     3
3     4
4   NaN
dtype: float64

I suspect this might make a nice API as well

s = pd.Series([1,2,3,4,'.'])
s.astype('float64', coerce=True)

maybe coerce=True|None|'raise' make sense

[TomAugspurger]

We don't want to overload .astype any further.

We have pd.to_numeric(s, errors='coerce'), and DataFrame.apply(pd.to_numeric, errors='coerce')

[tdpetrou]

I'd really like to see astype add the option for 'coerce'. Its quite confusing to have to be aware of the to_numeric function. Using it with apply is very ugly. It's just a single more option and a very good one in my opinion.