schemas/vegbien.sql: analytical_db_view: Renamed datasource's taxonverbatim to datasource_taxonverbatim to distinguish it from the other taxonverbatims that are joined on (parsed_taxonverbatim, accepted_taxonverbatim)

inputs/.NCBI/nodes/create.sql: Make genus (mostly) globally unique by removing kingdom Animalia, which has significant genus overlap with plants. This reduces the number of duplicated genera from 578 to 65 (determined with `SELECT name_txt, count(), array_agg(rank) FROM "NCBI".nodes GROUP BY name_txt HAVING count() > 1 AND 'genus' = ALL (array_agg(rank))`).

inputs/.NCBI/nodes/create.sql: Added foreign key on parent tax_id with covering index

input.Makefile: Staging tables installation: Added %/uninstall, %/reinstall to allow reinstalling individual tables

sql_io.py: put_table(): ensure_cond(): When adding the failed condition to the errors table, also include the original, untranslated condition from the DB schema in addition to the translation of the condition into the input schema

sql_io.py: track_data_error(): Fixed bug where errors whose column had no srcs (indicated by () ) were incorrectly being ignored. This affected NOT NULL exceptions where the column was not provided by the dataset.

sql_gen.py: If no cols had srcs, return [] instead of the [()] that itertools.product() would have returned

sql_io.py: track_data_error(): Support errors with no columns by inserting a single entry with column set to NULL

strings.py: Added join()

sql_io.py: mk_errors_table(): Made "column" column nullable, because some errors (such as check constraint violations) don't have any corresponding columns if its columns weren't provided in the input data