Make note of places where certain column/etc. names can't be used

[fedarko]

Summarizing this stuff

The user can't have sample metadata fields with the following names:

• Sample ID (except for the leftmost column, which we use as the index, of course)
• qurro_balance

Also, there can't be feature rankings (e.g. "Intercept", "Rank 1") with the following names:

• Feature ID (except for leftmost column)
• qurro_classification
• qurro_x
• LibraryID (if -gnps passed)
• anything in the feature metadata (now that More sophisticated feature metadata annotation #132 is done)

And there can't be feature metadata fields with the following names:

• Feature ID (except for leftmost column)
• qurro_classification
• qurro_x
• anything in the feature rankings
• qurro_trunc_feature_id (if -gnps passed)
• qurro_full_feature_id (if -gnps passed)

Looks like, when we call reset_index() on the sample metadata and rank data, an error will be thrown if a column with that name already exists. So Feature ID and Sample ID should be ok -- these cases won't fail silently (although it's a good idea to verify this in a test).

Also, my code checks for collisions w/r/t qurro_balance in the sample metadata, and between feature metadata/ranking field names (this is tested in test_df_utils.test_merge_feature_metadata!). So the main thing left to worry about is Classification not being overwritten, and then we should add in some tests that check ideally all of these cases to verify an error is thrown.

• Potentially prefix Classification with qurro_, like I did for balance.
• Check if Classification will be overwritten
• Add unit tests to verify that this sorta stuff will result in errors thrown
• Make note of these naming restrictions in the README

[fedarko]

This also applies when reset_index() is invoked, in the case that index is an extant column in the DataFrame for which this is called? Update: yeah, but we rename the index first so this isn't a problem.

[fedarko]

Similarly, if the user happens to have ranks in their input called "x" or "Classification" or "Feature ID", this will break.

[fedarko]

note for later -- there's also some stuff in the read_gnps_feature_metadata_file function that requires certain column name(s)? to not be present in the metadata file. that may or may not be temporary.