Currently, we hardcoded "linux/amd64" for the Docker images (both allocator and client). This is due to the Tensorflow backend of SLEAP. Since in the next release of SLEAP, there will be a shift to PyTorch, we should update the workflows to support multiple platforms.