Making Lebel assembly more flexible and customizable + bug fixes

[eyasayesh]

Proposed changes:

• Make selection of stories in the assembly customizable (c22cb1e)
• Make audio path optional and flexible. Current implementation hard codes it (e220f8a)
• Allow user to select path to brain response file, and eliminate the need to distinguish between surface and volume when loading file (1fc0ba2)
• Similarly, allow user to select path toward custom transcript file which contains all the processed stimulus stories. Additionally, change it to be a dictionary of dictionaries for each story instead of list of dictionaries (c1a6a46)
• Create a preprocessing helper module to support creation of custom huge brain_resp and processed_transcript files according to users desired stories (eb0e214)
• All other modifications are bug fixes that emerge when training an encoding model with volume data

Successfully trained an English 1000 model using all commits in this branch.