StringHashTable is really ObjectHashTable

[CarstVaartjes]

Hi,

pandas see strings as general objects, so it can be unicode or a BSON ObjectId or a list or something else. when it puts them in the hashtable, it gets the cstring name of the object (the name representation)
we should name this ObjectHashTable (a break with Pandas naming, I know, but really more to the point) and also have a real StringHashTable (we should check with bcolz how they define a string array)
Once that is finished we should look at accepting carrays, but that probably takes putting this code directly into bcolz carray_ext (as there is no pxd for carray_ext cimports)