The pipeline runs x times splitting based on the number of combinations of json

[suzannejin]

Current behaviour: The following json will end up with 3 combinations of parameters. For these 3 combinations, the splitting is always the same. However the pipeline will run 3 times for splitting, because we have 3 resulting jsons.
Desired behaviour: Run 1 times splitting, and 3 times noising.

{
    "experiment": "DnaToFloatExperiment",
    "interpret_params_mode": "column_wise", 
    "noise": [
        {
            "column_name": "dna:input:dna",
            "name": ["UniformTextMasker"],
            "params": [{"probability": [0.1, 0.2, 0.3]}]
        }
    ],
    "split": [
        {
            "name": "RandomSplitter",
            "params": [{"split": [[0.6, 0.2, 0.2]]}]
        }
    ]
}