[experiment] ty/layout: compute both niche-filling and tagged layouts for enums.

[eddyb]

This is something I've wanted to do for a while, since if we can get away with it perf-wise, it should allow us to always pick the better of the two, even when the conditions would be otherwise subtle.

[rust-highfive]

r? @petrochenkov

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

[eddyb]

@bors try @rust-timer queue

[rust-timer]

Awaiting bors try build completion

[bors]

⌛ Trying commit 13c200a with merge 9a594eaf4326233bc29fe01ece3e52c2a0406229...

[bors]

☀️ Try build successful - checks-azure
Build commit: 9a594eaf4326233bc29fe01ece3e52c2a0406229 (9a594eaf4326233bc29fe01ece3e52c2a0406229)

[rust-timer]

Queued 9a594eaf4326233bc29fe01ece3e52c2a0406229 with parent e82734e, future comparison URL.

[rust-timer]

Finished benchmarking try commit 9a594eaf4326233bc29fe01ece3e52c2a0406229, comparison URL.

[bjorn3]

There doesn't seem to be any perf difference between master and this PR.

[petrochenkov]

r? @RalfJung

[RalfJung]

I'm afraid I don't know this code nearly well enough to do reviews. I'd have suggested @eddyb until I saw they are the PR author. ;)

[eddyb]

This is only for the perf test 📈. If we did land this, we'd add a comparison between the two layouts, and pick the one that has smaller size, or more invalid values in the largest_niche (if they're the same size) - if somehow both of those properties are identical, we'd pick tagged because LLVM will likely handle it better.

[Muirrum]

@rustbot modify labels to +S-waiting-on-review -S-waiting-on-author

[RalfJung]

Assigning to @eddyb as it is just an experiment. r? @eddyb

[eddyb]

Thanks to @erikdesjardins, this was finished as #74069.