ENH: Add Index.fillna

[sinhrks]

Related to #6194. There seems no direct way to replace NaNs in Index?

I think a point is what types are acceptable via value.

• scalar: OK
• dict: NG (or accept {iloc : value} dict?)
• Series : NG
• DataFrame : NG
• http://pandas.pydata.org/pandas-docs/stable/generated/pandas.DataFrame.fillna.html

[jreback]

this should be a simple version, maybe with just value. the methods are already implemented via .reindex (e.g. pad/bfill/nearest). Don't need much functionanilty here (and in .dropna) but to fill (drop) the correct nan (NaT) type for that index and return a new one. This is actually a pretty simple method.

[jreback]

@sinhrks I think this is a preqreq for some other PR's.