Add Ref/RefMut try_map method

[GrigorenkoPV]

A more generalized version of filter_map, which allows to return some data on failure.

Safety

As far as I can tell, E cannot contain any 'b data, so it is impossible to duplicate the &'b [mut] reference into the RefCell's data.

Other than this E, everything is analogous to the already stable filter_map.

Try / Residual

I have considered generalizing this to use the Try & Residual just like #79711 does for array::try_map, but it does not seem to be possible: we want to essentially .map_err(|e| (orig, e)) but this does not seem to be supported with Try. (Plus I am not even sure if it is possible to express the fact that &U in Try::Output would have to have the same lifetime as the &T input of F.)

ACP

As far as I can tell, this is not mandatory, and the implementation is small enough to probably be smaller than the doc I would have to write.

rust-lang/libs-team#341

[rustbot]

r? @Mark-Simulacrum

(rustbot has picked a reviewer for you, use r? to override)

[Mark-Simulacrum]

ACPs are the recommended way to get the libs-api team to approve new unstable APIs. I've marked this as waiting-on-team which I believe also is in their queue.

(libs-api -- we may want to update the guidance on the linked docs to just require an ACP)

[m-ou-se]

As far as I can tell, this is not mandatory, and the implementation is small enough to probably be smaller than the doc I would have to write.

The motivation is an important part of the ACP. We do need to understand the motivation, ideally with clear use cases, before accepting new (unstable) APIs.

[Dylan-DPC]

@GrigorenkoPV any updates on creating the ACP? if you have done so, kindly add it to the pr description. Thanks

[GrigorenkoPV]

The ACP: rust-lang/libs-team#341

@rustbot review

[stegaBOB]

Thanks for your work on this @GrigorenkoPV! We would love for this method to be added. @Mark-Simulacrum is there anything else needed here to get it across the finish line? I'd be happy to put some time into this if I can help in any way.

[Dylan-DPC]

@stegaBOB this is waiting on the ACP getting merged, so we need to wait fro that and nothing is expected from author or reviewer about it.

[bors]

☔ The latest upstream changes (presumably #134052) made this pull request unmergeable. Please resolve the merge conflicts.

[GrigorenkoPV]

#133567 changed a place in rustc_interface where this method could be used and now it no longer fits, oh well.