May've been better, yes. My concern though is that with new syntax it'll get any traction, like, in the next millenium. Just looking at many other features stuck in either RFC or nightly stabilization limbo.
I'd love to have something like this that can be combined with deriveing a trait for the environment, or an equivalent mechanism. In Miri we need the equivalent of "a closure where the captured environment also implements a particular kind of visitor", and currently we need a macro for that which means we need to manually spell out what is being captured. I'd love to reduce that boilerplate.
For const_eval_select it'd be amazing to have a version of this for two-method traits (one method is const fn, one not), computing a single joint capture set for both of them.