[rustdoc] Box ItemKind::Impl to shrink the size of Item

[jyn514]

Helps with #79103. Builds on #79957 and should not be merged before. Eventually I want to calculate this info on demand (#76382) but that turned out to be quite difficult so I'm leaving it for later and slapping on this short-term fix.

This brings the size of Item and ItemKind from

[src/librustdoc/lib.rs:102] std::mem::size_of::<Item>() = 680
[src/librustdoc/lib.rs:102] std::mem::size_of::<ItemKind>() = 408

to

[src/librustdoc/lib.rs:102] std::mem::size_of::<Item>() = 608
[src/librustdoc/lib.rs:102] std::mem::size_of::<ItemKind>() = 336

Waiting to start a perf run until #79957 lands.

[rust-highfive]

r? @GuillaumeGomez

(rust-highfive has picked a reviewer for you, use r? to override)

[jyn514]

@GuillaumeGomez I plan to box a couple other variants too - would you prefer I do that here or in a separate PR? The only reason I'd separate it is to get more accurate perf numbers, but if we get a perf run on this I think we could compare later runs to the first one?

[GuillaumeGomez]

I'd prefer in other PRs yes. Like that we can compare more accurately and it makes reviewing easier for me. :)

[jyn514]

Another alternative I didn't think of earlier is to box the whole ItemKind. That would reduce the size a lot more without having to box each variant piecemeal.

[jyn514]

@GuillaumeGomez I made all the changes I planned to make so that I can compare them to the alternative approach in #80014. If we end up making this change I can split them into separate PRs, but I don't think it make sense to make any of these changes if 80014 gets merged, so I want to compare the final impact.

@bors try @rust-timer queue

[rust-timer]

Awaiting bors try build completion

[bors]

⌛ Trying commit 32b2b42 with merge 7fee63e6e685445eb5cb2757a4441b41db343277...

[bors]

☀️ Try build successful - checks-actions
Build commit: 7fee63e6e685445eb5cb2757a4441b41db343277 (7fee63e6e685445eb5cb2757a4441b41db343277)

[rust-timer]

Queued 7fee63e6e685445eb5cb2757a4441b41db343277 with parent 057937b, future comparison URL.

[rust-timer]

Finished benchmarking try commit (7fee63e6e685445eb5cb2757a4441b41db343277): comparison url.

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. Please note that if the perf results are neutral, you should likely undo the rollup=never given below by specifying rollup- to bors.

Importantly, though, if the results of this run are non-neutral do not roll this PR up -- it will mask other regressions or improvements in the roll up.

@bors rollup=never
@rustbot modify labels: +S-waiting-on-review -S-waiting-on-perf

[jyn514]

Max of -2.1% on instructions, max of -1.3% on max-rss. I think I would prefer to go with #80014 because it improves max-rss more.