Additional to #6898
Perform code coverage analysis against all iris tests
Untouched code is then either redundant or untested
Various parts of the code used to handle char<->string encoding, which is now done by "encoded datasets"
Notably:
fileformats.cf
fileformats._nc_load_rules -- mostly in helpers
fileformats.netcdf -- both loader and saver
Both load + save should now support the use of either string-arrays or char-arrays.
And all those behaviours should be tested.
But some obsolete code may have slipped though the net