Serialising sublists if they arise in ConstructAllGroups

[olexandr-konovalov]

At the moment, if sublists arise in the output of ConstructAllGroups, then the partial result is reported in the form of upper/lower bound on gnu(n), runtime and the pattern of length of sublists. In order to pick up the experiment from there, one should first rerun the initial call of ConstructAllGroups to reconstruct actual sublists.

I've checked that the test coverage of CodePcGroup and PcGroupCode in the current master branch is 100%. I suggest to serialise such sublists using CodePcGroup and PcGroupCode so that the groups contained in them could be analysed without calculation. Of course, recomputing the whole experiment has its own merit and will have to be done anyway, but having the list of group codes will give us more flexibility to pass them between different GAP instances, store etc.

(Of course, we could consider storing group codes for the whole collection, not only for sublists)