This is definitely an argument for a bigger standard library.
bincode is not the best at anything, which is why people should have, and many indeed did move on from it many years ago.
It's not particularly superfast (like speedy/borsh), doesn't cover zero-copy use-cases, isn't a standard (like cbor), and even serde support is gone in v2.
So it's actually yet another example of why those things SHOULDN'T be in the standard library.