Add pre-processing library to improve final results

It is often possible to improve results of the final processing by performing adjustements on the input image. We want to provide a helper library such that it is easy for users to try some pre-processing filters on the image in order to improve results.
As a starting point we want to implement:
- [ ] Contrast Adjustment and Document Binarization
- [ ] Margin Cropping (automatically detect a margin area and crop it)
- [ ] Margin OutCropping (automatically adding a margin to the document)
- [ ] Variable DPI rasterization
- [ ] Split and Stitch approach to automatically create multiple document from one very large image and infer it in parallel and stitch it back together.