bugfix: inputs/input.Makefile: postprocess must be run after cleanup rather than before because it depends on the cleanup having been performed.

this bug was not previously detected because this is only a problem when refreshing a datasource to data in the same format: this would attempt to run an existing postprocess.sql, out of order, instead of starting with no postprocess.sql as we usually do....

lib/sql_io.py: cleanup_table(): trim(): documented that this also converts character varying fields to text

bugfix: inputs/input.Makefile: $(dbExports): also need to put data.sql before clean_up.sql, etc. previously, this ordering had to be done by naming clean_up.sql, etc so they would sort after data.sql alphabetically, but it can be confusing to have to remember to do this. this fixes a bug in the CVS refresh where cvs.~.clean_up.sql was being run before data.sql, causing some private columns to have been deleted before the data was imported into the tables, creating a column mismatch error.

/README.TXT: To re-run geoscrubbing: re-create public-schema views that were cascadingly deleted: listed which views these currently are

/README.TXT: To re-run geoscrubbing: updated runtimes

bugfix: lib/sql_io.py: null_strs_str_default: removed "NA" because this is the abbr for a Spanish province (Navarra). this fixes the 2nd bug of #955, geovalidation duplicated rows.

lib/sql_io.py: cleanup_table(): debug-print null_strs

lib/sql_io.py: null_strs: made it customizable from an env var, since the same list of null_strs doesn't work for all datasources (see #957)

inputs/Cyrille_traits/Makefile: set custom $(null_strs) which handles "NA"

inputs/input.Makefile: pass make var $(null_strs) to invoked commands so it can be used by lib/sql_io.py