inputs/input.Makefile: %/install: Always generate a header.csv, even for CSV inputs with their own header. This will include the actual column names in the staging table, which may differ from their names in the CSVs (e.g. the addition of row_num). Note that header.csv is not included in the CSVs list itself, and will not override the header or dialect in them.

schemas/functions.sql: Added set_col_names()

schemas/functions.sql: rename_if_exists(): Also ignore duplicate_column exceptions, which are generated when a column is renamed to itself (as well as when two columns are renamed to the same place)

schemas/functions.sql: Added col_names(regclass), which unlike col_names(regtype) returns names in the order they are in the table

schemas/functions.sql: Added map_values()

schemas/functions.sql: map_get(): Fixed bug where can't use STRICT in EXECUTE INTO because there will sometimes be no match, causing a "query returned no rows" error

schemas/functions.sql: rename_cols(): Support any renames type with an -> operator

schemas/functions.sql: Added operator ->(regclass, text)

schemas/functions.sql: Added map_get()

schemas/functions.sql: table2hstore(): Made it STABLE instead of IMMUTABLE because the input table is not constant