[Feature Suggestion] Thresholding

Since SD does not work with alpha channel, thresholding is the only way we currently have to add it automatically.
Thresholding can also be helpful to detect and inpaint on specific use cases.
I am aware of preprocessor nodes, but neither segmentation nor depth is very precise when compared with thresholding for certain use cases and opencv already provides [multiple thresholding operations](https://docs.opencv.org/3.4/d7/d4d/tutorial_py_thresholding.html).

Meanwhile, here are my 2 cents: [mono-merge](https://gist.github.com/bmad4ever/ed93915829c18c3849a79997a2d33dc1#file-mono_merge_node-py). This allows you to merge two black and white images.