Bug: CategoricalReader assumes that the fieldtype metadata is not a tuple

[atbenmurray]

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