Added tnrs.py

tnrs_client: Factored TNRS request code into separate function tnrs_request()

inputs/VegBank/taxonimportance/map.csv: Documented that taxonimportance is not 1:1 with taxonobservation

mappings/VegCore-VegBIEN.csv: Removed unnecessary /_first/# suffix for multiple terms in the same _exists expression, because _exists() only checks whether its node is non-empty, and it does not matter how many child nodes it contains

schemas/vegbien.sql: taxonoccurrence: taxonoccurrence_unique_within_locationevent unique index: Fixed bug where locationevent_id needed to be enclosed in COALESCE so that the unique constraint also applies to rows with NULL locationevent_ids (there is no other unique constraint handling these rows)

README.TXT: Documented that if the row-based and column-based imports produce different inserted row counts, this usually means that a table is underconstrained (the unique indexes don't cover all possible rows). The inserted row count difference occurs because column-based import collapses empty table rows into one insert, while row-based import performs an insert of the empty row for each input row. Without a unique index to combine multiple row-based inserts, extra rows will be added.

sql_io.py: put_table(): Warn if inserting empty table rows

schemas/py_functions.sql: _namePart(): Fixed bug where it was returning the empty string instead of NULL

schemas/functions.sql, py_functions.sql: Added schema comment that functions must always return NULL in place of the empty string, to ensure that empty strings do not find their way into VegBIEN. Note that row-based import automatically removes empty strings because the intermediate values are stored in XML and our XML DOM traversing code auto-replaces the empty string with NULL. Column-based import, on the other hand, does not, because the intermediate data is stored in database temp tables instead of a DOM tree.

root map: Fixed custom public schema override to work with schemas lists that include public, by replacing public with the new public schema instead of just appending it