Clustering constructor is fickle

[jg-you]

Last issue I swear!
While testing the code for PR #29 , I noticed that the constructor of the Clustering object is very sensitive to the type of input given.

For example, when I construct a clustering from the following dictionary mapping clusters to elements

x = {0: [0,1,2,3,4], 1:[5,6,7,8,9]}
clu  = Clustering(clu2elm_dict=x)   

no warning are raised, yet a bunch of functions in the sim subpackage break down, because they expect a dictionary of sets.

I think that it would be easy to cast all iterables to sets at construction time, but I don't know whether this would affect other parts of the code?

[ajgates42]

Geez! I guess close to release I started to get sloppy ; ). I'll take another pass to clean things up.

[jg-you]

Nice!
While at it, it would probably be useful to support scikit style construction with clusterings declared as
[0,0,0,0,0,0,1,1,1,0,0,0] etc. I can put a PR in for that.