Reorganise datasets

[LucaMantani]

We should reorganise the datasets in the repo.

One idea could be to have the folders split by collider or similar. We should minimise repetition as much as possible too.

[jacoterh]

Yeah we could definitely do with some reorganising. We should

1. have the option to link the same theory to multiple datasets to avoid repetition
2. be consistent about how we name datasets: EXP_PROCESS_COM_CHANNEL_DIF/TOT_OBSERVABLE
3. get rid of L1 and make it run on the fly instead

Personally I don't see really what we gain by splitting by collider as long as we are consistent with how we name our files (see point 2)