[steffans]

Make sure to use registration keys for circular-dependency tracking instead of class names or factory source locations. This prevents false circular-dependency errors in those cases:

• A tagged singleton depends on an unresolved default singleton of the same type.
• Separate transient services use factory closures created at the same source location.

I added regression tests for both cases.

[github-actions]

Benchmark Results

Comparison of container-registration-cycle-tracking against 3.x (54b2aa8a9bf4d3cc9b555945f333da0ff4b263ee).

Open to see the benchmark results

Benchmark	Set	Mem. Peak	Time	Variability
ContainerBench(benchResolveDefinition)	-	4.104mb -0.00%	1.615μs -35.94%	±1.36% +156.74%

_{Generated by phpbench against commit 45628c9}