Docs now generate using Sphinx-gallery
Update conf.py and .rst files

Minor formatting changes.
Switched data_paths variable from dictionary to list

value as size of BE_train_wave

Due to different requirements for the GLineBE translator to use 'AI_wave', 'BE_wave_AO_0', 'BE_wave_AO_1', 'BE_wave_train', 'BE_wave' etc. for the given dataset. We are creating another filter that can be used as per our requirements.
Some more changes might be pushed into it based on the type of data and vectors we want in our HDF5 file later on.

Test if it is able to install properly next after creating the file.

Use the new get_attr in function

Cades dev

About pycroscopy, why it exists, why it is needed, and how it is implemented at a high level

Added the passing of the number of points to use in the SHO guess

num_points should no longer be ignored when passed in options of do_guess

Fixed label creation if not provided
Minor formatting changes to be_utils and be_loop

Added df_utils to imports

Create a new hdf5 file after deleting the old one.

Added a shape attribute to the object
Added a __getitem__ method to the class

Updated the parse_dm3 file.
The StringIO imports should now work on both python 2 and 3.
The objects read from the binary file are now decoded as utf-8 when possible.  If decoding fails, the error is ignored and the file read procedes without crashing.

Ensure that the batch size is an integer when cleaning in batch mode

Now require matplotlib>=2.0.0 to ensure we have the new default viridis colormap

Added a helpful (hopefully) message if the channel type is not found so that the user can add it if needed.
Some attributes are stored as np.object_ rather than np.bytes_.  Now checking for that as well.