GPU acceleration of the reproject package

This is a copy of a question I posed on the slack channel opened up to all users

Hello! I am working on speeding up the reproject package using the GPU. I've already updated the pixel to pixel functionality for a 30% reduction in computation time for this algorithm. I'm going to be working on updating the other functions in the package to run on the GPU. I wanted to know if anyone has already done this or if someone is currently working on something similar. According to the Roadmap (or at least this is how I understood it), there is a need for someone to work on this type of implementation. Would it be possible to talk to anyone on the dev team about this? I'm going to be working on this acceleration either way, so I'd like to be able to contribute if the community thinks it would be helpful

Start by reviewing the reproject package's pixel-to-pixel functionality and the other functions proposed for GPU execution, along with the roadmap context mentioned in the issue. The issue does not define a target entry point, supported GPU approach, tests, or completion criteria, so clarify the scope with maintainers before starting.

Written by the indexing model from the issue text.