option to remove overlap duplication

[eroux]

there should be an option in the ocr import pipeline to detect and remove duplicates as in

https://github.com/OpenPecha/OCR-Pipelines/issues/8

This should be relatively (!) straightforward by looking at pixel coordinates within a line and either:

• remove everything that has a big enough overlap (this is the best but more complex option)
• remove everything that goes back horizontally

[eroux]

the intervention should be around here @ta4tsering do you want to give it a try?

[ta4tsering]

okay I will look into it