[HistFactory] add support for sample weights

[blinkseb]

The weights expressions must be defined in the python configuration, through the sample_weights dictionnary. The key is an identifier, and the value the expression of the weight.

In the sample json, one can use one of the key defined in the python configuration for the new sample-weight option. The weight evaluated from the expression will be used to scale all the histograms. If set to none or not specified, a default weight of 1 is applied.

Example:

In python, one can do

sample_weights = {
    'cluster_1': 'getWeightForCluster(1)',
    'cluster_2': 'getWeightForCluster(2)'
}

In the sample json, you can apply a given sample weight using the sample-weight option. It can be either cluster_1, or cluster_2.

{
    "TT_TuneCUETP8M1_13TeV-powheg-pythia8_MiniAODv2_v1.0.0+7415_TTAnalysis_12d3865": {
        "tree_name": "t",
        "files": [],
        "db_name": "TT_TuneCUETP8M1_13TeV-powheg-pythia8_MiniAODv2_v1.0.0+7415_TTAnalysis_12d3865",
        "sample-weight": "cluster_1"
    }
}

[blinkseb]

PR updated. The sample weight is now applied only on MC and disable on data.

[swertz]

Thanks! Let's reweight those babies!