COMP0016 Team 15

I have now added feedback and records of downloads to the backend, when a user downloads a model, this is recorded in the database, and then users can give feedback for models they have downloaded. I have also added in delete functionality, so users can delete models and datasets.

An issue I ran into was that since I have a computer using ARM64 architecture and the Linode server uses AMD64, making these models slow or unusable for people with different machines. This inspired me to add multi architecture support. I attempted this with buildx, but this only allows output to docker hub, which will not work for our closed system. So instead there are now two docker files in model_gen, one which builds from an ARM64 python image, and one for AMD64. An then instead of .tar files stored in /models directories of the model name are stored, with each one containing two tar files, one for the ARM64 image, and one for the AMD64 image.