Added tables for chi2 and phi by process

Created by: wilsonmr

As per the email thread

This is possibly reproducing in some way a function which Rosalyn and Cameron are already using to construct the theory covmat by process. I thought I should make the PR to make sure I'm not doing something stupid - and also so that the corresponding runcards for the tables in the paper will eventually work on master.

@scarrazza perhaps this is relevant/should be kept in mind with #476 where ideally the data specification was a flat list of data and then I wouldn't have to do the hack I'm doing here where I export the grouped datasets under the key experiments which causes problems if the user actually specifies experiments at the runcard level.

I just realised that these two functions (*groupby_process and *_groupby_experiment) could be merged into one function where you can declare the grouping at runcard (maybe with the default one way or the other) and then all of the relevant plots and tables would be flexible and config.py would look less messy with duplication of functionality - I will look at this later