if a variable is uncategorised *and* has no cleaning/processing rules set -
now the decision is based purely on whether or not a variable is
categorised. Previous logic was far too complicated.

Mnt/warning

See merge request !76

possibility of an empty vid list being passed to a processing function

variables) which are not present.

to support binariseCategorical. All this stuff needs to be re-written and
cleaned up.

generate/return broadcast indices. Instead, missing vars are filtered out
within runProcess, and a warning emitted to that effect.

Rf/newlines

See merge request !75

non-[numeric|datetime] columns

Bf/expression

See merge request !74

arrays and not lists

https://github.com/pandas-dev/pandas/pull/40180)

float and int appears to always return not equal. Solution is to cast values
to be replaced to the appropriate dtype

used to be possible to assign a np.datetime64 to np.nan (although it doesn't
make any sense, thinking about it now). If dtype is datetime, use a fill
value of np.datetime64('NAT')

not present

no relevant columns

CI: make sure funpack is on pypath

See merge request !73

CI: job tags

See merge request !72

Rf/dupe columns

See merge request !71