Make parallel iteration test less sensitive to timing

[ngoldbaum]

Fixes #5752 (I think)

I haven't been able to reproduce the failure on my Macbook. I suspect that this is a timing issue though, so I rewrote the test to improve synchronization between the worker threads and also make it so both threads exit the loop as soon as the borrow error happens.

[codspeed-hq]

Merging this PR will degrade performance by 13.62%

❌ 1 regressed benchmark
✅ 99 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	format_simple	185 ns	214.2 ns	-13.62%

---

_{Comparing ngoldbaum:fix-parallel-iter-test (a77d24c) with main (8f0af50)}

[davidhewitt]

Thanks for debugging, this makes sense to me!