Fix Possible Multi-scale Prediction Overwrite in Post-loop Assignment

[nya-a-cat]

Hi Zhao,

This line appears to use overwrite assignment to reassign the return value of scale_process to the prediction variable. Prior to this, within the loop, the code was already using prediction += scale_process(...) to accumulate predictions from different scales.

I am wondering if this overwrite assignment after the loop is intentional? My understanding is that it might overwrite the previously accumulated multi-scale results, and the final prediction would only retain the prediction from the last scale. If so, I am unsure if this aligns with the intended behavior of multi-scale testing.

Best.

Originally posted by @nya-a-cat in 5e5a0ba