inputs/GBIF/_MySQL/MySQL.data.sql.run: documented that import_resume_pos() takes 6 min to run, with 37 iterations

added inputs/GBIF/_MySQL/MySQL.data.sql.run, with helper functions for resuming the import to MySQL from where it left off. this is very useful if the import is interrupted for any reason, because otherwise, the entire import would have to be run again from the start, taking 40-50 hours. import_resume_pos() uses new binsearch() to find where in the file the import left off, based on which pkeys have already been imported. (GBIF pkeys are unfortnately not in any order in the input file, nor are they in insertion order in the imported table, because MySQL instead clusters the table by the pkey. this necessitates a much more complex solution to resuming a partial import.)

lib/sh/binsearch.sh: binsearch(): also echo_vars the iter_num, to track how close binsearch is to finding the value (it will always take the same # iters, log2(max - min) )

lib/sh/binsearch.sh: binsearch(): also echo_vars the min/max so these can be used as shortcut inputs if binsearch is run again

bugfix: lib/sh/util.sh: caching: cache_key for function inputs: need to use `declare -p kw_param` instead of "$kw_param" because declare accepts a param name, not value`

lib/sh/binsearch.sh: binsearch(): doc comment: fixed typo in "truncates"

bugfix: lib/sh/util.sh: func_override(): need to match shortest _* suffix instead of longest in case the function being overridden itself contained _

bugfix: lib/sh/util.sh: file_size: Linux: need % in %s

lib/sh/db.sh: mysql(): added $data_only flag which enables --skip-column-names and $output_data

bugfix: lib/sh/util.sh: file_size: need to use --format instead of -f on Linux