Simplify WaitGroup implementation

[Veykril]

No description provided.

[netlify]

✅ Deploy Preview for salsa-rs canceled.

Name	Link
🔨 Latest commit	1b15e41
🔍 Latest deploy log	https://app.netlify.com/projects/salsa-rs/deploys/688f17b5bfc9f40008bad483

[codspeed-hq]

CodSpeed Performance Report

Merging #958 will not alter performance

_{Comparing Veykril:veykril/push-rqyrkyttxpol (1b15e41) with master (c3f86b8)}

Summary

✅ 12 untouched benchmarks

[MichaReiser]

Is there any other benefit than code size reduction. It seems codspeed doesn't like it at all.

[Veykril]

No there isn't. I was curious about the perf impact, I agree this isn't great. Checking if we can actually simplify our impl now

[Veykril]

Hm, we could get rid of the additional Arc as well and re-use the Zalsa arc but that makes the code a lot more ugly (thanks to having to deal with drop order), so this is the best I can come up with. Trading an unwrap with unsafe solely due to polonius borrow checking not being a thing yet :(

[MichaReiser]

@ibraheemdev could you take a look at this PR?

[ibraheemdev]

Suggested change

	if let Some(zalsa) = Arc::get_mut(&mut self.handle.zalsa_impl) {
	// SAFETY: Polonius when ... https://github.com/rust-lang/rfcs/blob/master/text/2094-nll.md#problem-case-3-conditional-control-flow-across-functions
	break unsafe { mem::transmute::<&mut Zalsa, &mut Zalsa>(zalsa) };
	}
	if Arc::strong_count(&self.handle.zalsa_impl) == 1 {
	// SAFETY: The strong count is 1, and we never create any weak pointers,
	// so we have a unique reference.
	break unsafe { &mut *(Arc::as_ptr(&self.handle.zalsa_impl) as *mut Zalsa) };
	}

This should avoid the Polonius issue and is a bit cheaper. There is a little question about writing through Arc::as_ptr, but it is valid in the current implementation and nothing is really blocking it being guaranteed (see rust-lang/rust#104337).