Implement GroupBy __len__ to return number of resulting groups

[bburan]

Right now there is no way (to my knowledge) to get the number of groups that the DataFrame.groupby() method creates. For example (using Pandas 0.4 dev):

from pandas import *; from numpy.random import randn
data = { 'A': rand(4), 'B': rand(4), 'C': (0, 1, 1, 1)}
df = DataFrame(data)
grouping = df.groupby('C')
len(grouping)

TypeError: object of 'DataFrameGroupBy' has no len()

This seems counter intuitive considering iter is implemented. Having the len() method would be very useful if we need to know the number of existing groups prior to iterating through the groups (e.g.. one use case would be to determine the number and layout of subplots, one for each group, in a Matplotlib figure).

Right now, I typically just do:

grouping = list(df.groupby('C'))
n = len(grouping)
# do something with n

for key, frame in grouping:
# do something

(updated to fix formatting)