Ben Murray

Docstrings for exeteracovid processing classes

This would enable optimisations for long stretches of invalids indicating no corresponding value exists to join

Although it isn't considered high priority at the moment, there are still potential scenarios where a user might want to import string-based numeric representations and convert them to numeric formats...

For categoricial fields, there should be an option to flag unrecognised values rather than raising an exception and stopping import. It would nice to report that unrecognised values were found....

However, it typically is. Other readers work around this by performing a split on comma following by selection of the first element. CategoricalReader did not get this, for some reason

Sort for fixed string field should yield major improvements over numpy. Cast of underlying data to char array is key.

We need a fast fixed string get_spans in cython

We need a fast safe_map_valid for fixed strings

We need a scalable version of apply_indices for fixed string fields in cython. It should perform the operation in a chunked fashion

We need a scalable version of apply_filter for fixed string fields in cython. It should perform the operation in a chunked fashion