Reformat column name handling

[dltamayo]

Theres hardcoded column names all over the place in this code, consider factoring them out into 1. named function arguments with default values 2. a constants.py file containing the default values

Otherwise when one of these column names changes, or someone shows up with data with different column names, or even lowercased column names, there will be a huge headache.

Also if there are hardcoded column names, somewhere we need to check the input data and give user a readable error message like "expected these columns .. got these columns .." to make it possible for people to correct their data on their own

Originally posted by @qwwqwwq in #42 (comment)

[dimalvovs]

Was this done in #42 ?

[dltamayo]

Not implemented yet, #42 referred to data directories, this is about columns in the input TSVs

[dltamayo]

Implemented in #67
Only required metadata columns now are sample and file; everything else can be specified as parameters for the sample-level notebook.