sql.py: mk_insert_select(): embeddable: Fixed bug where the function may do different things when run, because the function (and other statements whose cached strings depend on the function name) may be run after the function definition would have changed, by versioning the function name and using CREATE FUNCTION instead of CREATE OR REPLACE FUNCTION so that its definition never changes

sql.py: Parse "function already exists" errors as DuplicateFunctionException

sql.py: mk_select(): joins: Fixed bug where join_not_equal did not do what it was designed for, which is filtering out matches of the join condition (before the bug fix, it effectively did a cross join with matching rows excluded, causing duplication of rows). Renamed join_not_equal to filter_out to reflect its intended use. Support table-scoped column names in the WHERE conds list.

sql.py: put_table(): Fixed bug where ORDER BY column needed to have table0 name prefixed (if it didn't already have a table name), to avoid ambiguous column references

sql.py: mk_select(): Fixed bug in joins where right_col had the table name prepended before it was copied for use with a different table name in join_using and join_not_equal

Mapped some unmapped fields in DwC inputs

Added mappings/for_review/DwC2-VegBIEN.specimens.fields.csv

db_xml.py: put_table(): Fixed bug where didn't commit right after inserting node, but instead waited until children with fkeys to parent (independent of the node itself) were inserted

sql.py: put_table(): insert_(): Use insert_select() instead of run_query_into() if new option pkeys_table_exists is on

sql.py: mk_select(): Support joins with !=