Date formatting option in to_csv?

[wesm]

http://stackoverflow.com/questions/13999850/how-to-specify-date-format-when-using-pandas-to-csv
http://stackoverflow.com/questions/15651527/how-to-stop-python-pandas-from-adding-000000-to-every-date

[jreback]

can easily support in new to_csv via say date_format (as the DatetimeBlock can handle analagously to float_float by the FloatBlock), pushing to 0.12

Also should add a couple of methods (in format.py?) to handle the various cases for passed formatters

e.g.
%.2f
lambda x: "mycool float: %.2f" % x

This needs to be uniform across all IO apis, and flexible enough.

For example, float_format doesn't support new style python formatting
(no thousands seperator for example). Formatters is a better concept,
but not as concise.

We need to find a better way to this.

fwiw, modifying the frame and then exporting is actually a nicer
seperation of concerns. formatting arguably doesn't belong in
to_csv.

[jreback]

easy enough to have a class that can accept multiple input types (new style formatting,old as strings lambdas), then provide an api to the funcs that need it

yeah, I'm for doing the formatting as a seperate stage using a delegate class,
and then having IOs use that across the lib. not sure how that would work.
0.13?

[jreback]

also need a top-level option for nat_rep='NaT' (for putting NaT) instead of an empty string

[jreback]

closing in favor of master issue #4668