Consider removing column categories in digest schemas

[alyssadai]

The differentiation of GLOBAL_COLUMNS vs. PIPELINE_STATUS_COLUMNS (or ASSESSMENT_SPECIFIC_COLUMNS in the phenotypic schema) is partially an artifact from when different pipeline steps were represented as separate columns. These categories may not serve much of a purpose now that we have pipeline_step and status.

One option would be to remove these top-level keys entirely from the schemas. Alternatively, we could explore other categories, if there is another grouping that would be useful for digest file generation.

[michellewang]

I think it makes sense to remove the top-level keys now that all column names are fixed