EncoderNormalizer Question

Would someone please explain what the EncoderNormalizer does using the Stallion tutorial as a concrete example?

I see in the documentation states "this transformer will be fitted on each encoder sequence separately", but I don't know what "fitted" and "each encoder sequence" means.

With respect to the Stallion tutorial, is each encoder sequence each agency/SKU pair?

Does fitted equate to min/max scaling where each feature dimension is scaled by the following:
scaledFeatDim = (FeatDim - min(FeatDim) ) / ( max(FeatDim) -min(FeatDim) )

Or does fitted mean mean subtraction, unit standard deviation?
scaledFeatDim - (FeatDim - mean(FeatDim) ) / std(FeatDim)

Or does fitted mean something else entirely? I would really love an equation if possible, pretty please.

As always, I'm super grateful for this library, and all the hard work that was put into it.

Start with the linked EncoderNormalizer API documentation and the Stallion tutorial. Determine what “fitted” and “each encoder sequence” mean in that example, identify the normalization equation used, and update the relevant documentation so the explanation is concrete and complete.