Run the two models and turn their output into a finished patch

[KurbyDoo]

Background

Once the mock models and elevation encoding are complete, we can wire the two together and apply them to input patches instead of the blur function

Description

• Create a function called clean_patch that implements the same interface as the blur function
• Within this function, load the two models using load_model and wire the input and output grid to each other
• The core model should output two grids, a latent map and a low resolution grid, the latent map gets passed to the decoder
• Load the elevation encoding module and pass it the decoder output and low resolution grid
• Return the output

Testing:

• Assert input and output size are the same after the pipeline executes
• Assert the same patch run twice gives the same output