inputs/.TNRS/schema.sql: tnrs_populate_fields(): removed no longer needed casts of *_score to double precision

inputs/.TNRS/schema.sql: tnrs: *_score: changed type to double precision because these fields are always floats. this also avoids the need to manually cast them to double precision each time they are used.

lib/tnrs.py: HTTP requests: rewrapped lines

lib/tnrs.py: updated HTTP requests to match current web app

bugfix: lib/tnrs.py: download_request_template: changed dirty to true (to match the current web app), which is apparently needed to apply the source_sorting setting to the downloaded TSV in addition to the GUI results

lib/tnrs.py: retrieval_request_template: turned source_sorting back off, because it causes any match from the first source to always be used, even if it has a lower match score than the match from the other source. (Brad confirms that this should be off.) I think we had this on originally to ensure that only Tropicos results were used when available, rather than USDA when it was a better match. * note that due to a bug in the web app, this change will not actually be effective, because the source_sorting option is only applied to the GUI results, not the downloaded TSV. *

inputs/.TNRS/schema.sql: tnrs: Name_number: changed type to integer so it would sort numerically

inputs/.TNRS/schema.sql: added pkey on Time_submitted, Name_number

inputs/.TNRS/schema.sql: changed Name_submitted pkey to a unique constraint to allow adding a pkey on Time_submitted, Name_number instead

inputs/.TNRS/schema.sql: Time_submitted, Name_number: added NOT NULL constraints so that they can be used in a unique constraint