Pre-RFC: Deprecation and separation of the dead_code lint

skade:

I want to add that "dead code" is a common industry term that e.g. also finds its way into coding standards. "rustc can detect dead code by itself accurately" is a pretty big selling point and the feature having that name is really useful.

The problem is, as shown by my example above, not all dead code that rustc detects triggers the dead code lint. So the dead_code rustc feature is not the same thing as "the rustc feature that detects dead code".