Make IntArrayFormatter and all upstream users able to specify formatting for integer types

[blais]

This was found while attempting to add thousand separators (commas) on a Series.to_string() call. It looks like you can specify the format for floats, but not for integers.

It's impossible right now. I could transform the Series instance into a FP representation, but that is the wrong thing to do.

[ghost]

Yes, float_format isn't as general as it ought to be, and only supports
old % style formatting.

For now, maybe just:

s=pd.Series(range(1000,1025))
pd.Series(map(lambda x: "{:,}".format(x),s.values))
Out[11]: 
0     1,000
1     1,001
2     1,002
3     1,003
4     1,004
5     1,005
6     1,006
7     1,007
8     1,008
9     1,009
10    1,010
11    1,011
12    1,012
13    1,013
14    1,014
15    1,015
16    1,016
17    1,017
18    1,018
19    1,019
20    1,020
21    1,021
22    1,022
23    1,023
24    1,024
dtype: object

[ghost]

related #2502

[hayd]

closing in favor of master issue #4668