GitHub is for source code hosting. Not tens to hundreds of MB binary hosting. Put your binaries in s3 and link to them from your project if needed.
They probably determined the excess bloat and strain on their infrastructure from git cloning huge binaries outweighed the benefits so they are removing the feature.
The downloads aren't stored in git. They're stored on Amazon S3. Every download link redirects to cloud.github.com, which is fronted by Amazon CloudFront.
Which, conveniently enough, is exactly what they want us to do ourselves now. I already pay GitHub to do that for me, how is it a benefit for me, the paying customer, to do that myself by hand?