A large scale Cropsar 2d testing in CDSE

[HansVRP]

The ticket is for testing and validating the output with the CropSAR experts. Currently the following changes are observed as of 20/12/24.
The image on the left is the initial cropsar with old Biopar model and using the old masking function.
The image on the right os the updated cropsar with new Biopar model and using the latest masking function.

[HansVRP]

• Can we compare once more with the terrascope benchmark from Hazan?
• Could you create an epic for cropsar where we keep track of all seperate ongoing issues?

--> Ideally both produce the exact same output