Run fixpoint per strongly connected component (no sync table)

[MichaReiser]

Run a single fixpoint iteration for strongly connected components
to avoid exponential runtime for deeply nested cycles.

TODO:

• Documentation
• Investigate perf regressions
• Test that demonstrates the "collapsing" of nested cycles
• Review if there are any changes necessary to maybe_changed_after
• Investigate why cycle_nested_deep_conditional_changed::the_test sometimes hangs or panics

[netlify]

✅ Deploy Preview for salsa-rs canceled.

Name	Link
🔨 Latest commit	6e42e23
🔍 Latest deploy log	https://app.netlify.com/projects/salsa-rs/deploys/68e817fcb972e600089d6c6f

[codspeed-hq]

CodSpeed Performance Report

Merging #995 will degrade performances by 12.14%

_{Comparing MichaReiser:fixpoint-scc (6e42e23) with master (8b0831f)}

Summary

⚡ 1 improvement
❌ 5 regressions
✅ 7 untouched

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

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
❌	amortized[Input]	2.1 µs	2.4 µs	-10.33%
❌	amortized[InternedInput]	2.1 µs	2.3 µs	-9.33%
❌	new[Input]	9.6 µs	10 µs	-4.06%
❌	new[InternedInput]	4.3 µs	4.7 µs	-7.78%
❌	converge_diverge	125.4 µs	142.7 µs	-12.14%
⚡	converge_diverge_nested	200.8 µs	154.3 µs	+30.11%

[MichaReiser]

Assuming I'm getting this to work. One interesting optimization for memo.block_on is to have a DependencyGraph::all_cycles(queries) method instead. That's much cheaper than iterating over all memos (and only requires aquiring a single lock)