Once $minMaxScaler implementation complete, add "minMaxScaler" as an option to 'inputNormalization'.

To implement this, optionally add a $setWindowFields stage after the $addFields stage that computes the un-normalized score.

So the score adding stage(s) will go from:

{$addFields:
  {<input_pipeline_name>_score:
    {$multiply:
      [{"$score"}, <pipeline_weight>]
    },
  }
}

to: 

{$addFields:
  {<input_pipeline_name>_score:
    {$multiply:
      {"$score"}, <pipeline_weight>]
    },
  }
},
{$setWindowFields:
  {
    sortBy: { 
      <input_pipeline_name>_score: -1 // low scores first 
    }
    output: {
      <input_pipeline_name>_score: { 
        $minMaxScaler: {input: "$<input_pipeline_name>_score"},
    }
  }
}

I haven't actually checked if its possible to make the input and output fields the same (effectively update / overwrite a field) with the $setWindowFields stage. However, if we get an error doing this, we should be able to call the unnormalized score (output of $addFields) one field, and then the normalized score (output of $setWindowFields) the final score field - and then after add a $project after that removes the unnormalized score.