Additional to #6898 * review+tweak any public api + docstrings * write a detailed subsection within the [netcdf fileformat description](https://scitools-iris.readthedocs.io/en/v3.14.1/further_topics/netcdf_io.html) * explain choices + controls, for both load+save * explain "choice" to have either char-arrays or string-arrays in cubes/coords * simple examples of loading * attachment of `_Encoding` attributes to control saving * explain/show how to correct ["overlength when encoding" error](https://github.com/pp-mo/iris/blob/ef11375e99ccfff6bf780cdef732a98e2a3ffc2a/lib/iris/fileformats/netcdf/_bytecoding_datasets.py#L97-L104) * (i.e. use `numpy.astype` to lengthen the string dim created)