refactor and add nilreason logic

[rshewitt]

related to #4924

reference to some improvements made but i'll list them here anyways as a formality.

1: most top-level modules like module_metadata, module_iso19115_2, and module_metadata_info always return a hash. low-level modules like module_date, module_citation, and module_responsibility occur nested within a document and either return nil when the root element can't be found or the hash as it could be processed.
2: i throw a structure error message either when the input document is empty or it's not well formed xml. all other warnings are validation messages.
3: this is implemented. each module will try to gather as much information as possible without returning nil on the first warning.
4: nilreason check is implemented for required elements. it's not implemented for optional.
6: this relates to nilreason.

here's some general improvements

• changed tab width to 2 spaces instead of 3.
• removed certain readers & tests from linting. some of them weren't originally included and they produced errors but we only care about the _2_datagov branch for now.
• in order to make sure the nilreason logic works as intended i've broken required element xpaths down to 1 element instead of 2 ( i.e. gmd:westBoundLongitude/gco:Decimal to gmd:westBoundLongitude ). the reason is that gmd:westBoundLongitude is what's checked for the nilreason.

full translate test

• i created this test for validating documents in production. the documents from that query can change which is why i didn't commit this.

[btylerburton]

Nice work on this huge effort. LGTM!