browsing refactor

[anwarnunez]

At the moment, we're overloading __dir__ and __attr__ in order to return python objects that allow tab completion. Unfortunately, this causes issues when the s3 object name created starts with a number. (Python only allows object names to begin with a letter or an underscore, and we need to create objects to browse).

At the moment, we will avoid the problem by pre-pending NUM_. The full actual s3 object name is always accessible under _fullpath in a browser object.

In the future, we should use this: https://github.com/h5py/h5py/blob/master/h5py/ipy_completer.py

>>> import cottoncandy as cc
>>> cci = cc.get_interface('my_bucket')
>>> cci.ls('/myfold<TAB>')
myfolder1    myfolder3 
myfolder2    myfolder4
>>> cci.ls('/myfolder2/file<TAB>')
file1    file3 
file2    file4