change: Split Behaviour

[mathysgrapotte]

Currently, split_csv is expected to work with a sub-config. Which is different from the overall nextflow design (split once - perform experiments after).

Two possibilities :

implement sub-splits, would look like this :

full_yaml -> yaml_with_split_1 ; yaml_with_split_2
yaml_with_split_1 -> subconfigs
yaml_with_split_2 -> subconfigs (pipeline ran twice)

re-run split with every sub-config (could be expensive).

Only allow one split option