Strategy for handling cell_methods

### Something else
General approach to variables with `cell_methods` in CSET, i.e. time-mean quantities, or max-in-hour temperature.
Our standard approach is to filter these out in recipes and ignore them, looking for cell methods [].
Two problems that we will rapidly face;

- RAL4 and screen_temperature max/min in hour; which will need handling (including appropriate ways to specify in GUI names). We can probably relax the cellmethods constraint do some more clever filtering in read.py for wanted variables.

- FastNetUK, and general precipitation aggregation to 6 hourly or similar, as done in the RES, and for broader model/obs comparisons

Start by reading the recipe handling and read.py logic that filters for empty cell_methods, then inspect how variable names are exposed in the GUI. Work through the RAL4 and screen_temperature hourly extrema cases and the FastNetUK or broader precipitation aggregation case; done should define a clear strategy for selecting and naming these quantities.

Written by the indexing model from the issue text.