sql_io.py: put_table(): Setting pkeys of missing rows: Fixed bug in column-based import where when input rows match multiple output rows in one of this iteration's input tables, the into table's pkey constraint is violated because full_in_table contains multiple entries for an input pkey

schemas/vegbien.sql: place: Added index on matched_place_id to facilitate looking up places by matched_place_id in place_set_canon_place_id()

schemas/vegbien.sql: Added geoscrub_input derived table with associated view. Build geoscrub_input as part of make_analytical_db().

sql.py: Added table_has_pkey()

sql.py: table_pkey_col(): For PostgreSQL DBs, use pg_catalog via index_cols() and table_pkey_index(), in order to use the search_path to look up the table. This fixes a bug where the pkey would be selected from information_schema.table_constraints in random order, and this order sometimes returned the corresponding table in the public schema but sometimes in other schemas, such as VegBank. This became a problem now that VegBIEN has a place table, which conflicts with VegBank's place table. (Most other VegBank tables that are mapped to have been renamed in VegBIEN.)

sql.py: add_pkey_or_index(): Use new add_pkey_index()

sql.py: Added add_pkey_index()

inputs/import.stats.xls: Updated import times

import_times: Use $'' quoting to expand tab, in order to also work on Mac OS X

schemas/vegbien.sql: place: Added place_set_canon_place_id_on_*() triggers, analogous to taxonlabel.taxonlabel_2_set_canon_label_id_on_*()