Hmm. That's pretty large. I have to say that I don't feel great about this. Particularly since that rule is hardly a popular one -- nor the most crucial. Perhaps it was too ambitious to imagine changing it and instead it should remain a lint forever (it is, in its own way, a kind of lint anyhow). (cc @aturon)
I think we need to drill more into the process for these warning -> error transitions. For example, here, because this warning was so old, I don't see that we did any crater runs of impact (can that be true? I feel like I do remember some investigation?). However, for https://github.com/rust-lang/rust/pull/34982, we did a crater run, and then I made a point of ensuring that every affected crate was at minimum notified and, if possible, had a pending PR fixing the problem. I believe that for https://github.com/rust-lang/rust/pull/34923 we were similarly diligent.