ImageEstimators.ExtractPixels using an array for offsetImage and scaleImage

Hello,
for this moment, when we use ExtractPixels, we could provide only a single number for offsetImage and scaleImage, which will be applied for all RGB channels. But sometimes, as for [this model](https://github.com/onnx/models/tree/master/ssd), we should use different normalization across the channels. Perhaps, this functionality could be added in the future?
The same proposition for ImageEstimator.ResizeImages, perhaps, different interpolations for ResizingKind could be added as in PIL python library?