obligations_for_self_ty: skip irrelevant goals (recompute sub_root from stalled_vars)

[inq]

Supersedes #146759. Since the original PR is having conflict now, I've rebased.

This includes 2 commit

1. the original one with rebase
2. above that, I've changed:

if stalled_on is Some: ignore sub_roots -> walk stalled_vars -> re-evaluate sub_root 
else: same with this PR

Why:
cached sub_roots can be stale if there is a sub-unification merge after stalling (lcnr proposed on the original PR)
so re-evaluating when reads ---> so more conservative

Testing:

• next-solver / traits tests
• we can follow after merging Implement a new flag -Zdisable-fast-paths in trait solving #156172 with -Zdisable-fast-paths

Style question:

In my preference, inside filter, I used filter_map - matches, but do you prefer if-let chain?

Original author: lcnr #146759

r? @lcnr 

[rustbot]

changes to inspect_obligations.rs

cc @lcnr

Some changes occurred to the core trait solver

cc @rust-lang/initiative-trait-system-refactor

[lcnr]

please shallow_resolve the stalled_var here and if it is not an inference variable, always return true for now.

View changes since the review

[lcnr]

also, please add a comment explaining that we are conservative here and want to handle cases where a goal is no longer stalled

[inq]

added! thank you!

[lcnr]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: 1a8e5b7 (1a8e5b75d2437597d5622146fe1d722a41e373c6, parent: 4feb7221f4d445120a5061b16ce7222adbfdf6f6)

[lcnr]

r? BoxyUwU

given that it's largely my PR, let's give it to somebody else to review

[rust-timer]

Finished benchmarking commit (1a8e5b7): comparison URL.

Overall result: ✅ improvements - no action needed

Benchmarking means the PR may be perf-sensitive. It's automatically marked not fit for rolling up. Overriding is possible but disadvised: it risks changing compiler perf.

@bors rollup=never
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-17.6%	[-51.4%, -0.5%]	9
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

Results (primary -2.4%, secondary -67.8%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-2.4%	[-2.4%, -2.4%]	1
Improvements ✅ (secondary)	-67.8%	[-90.8%, -0.5%]	4
All ❌✅ (primary)	-2.4%	[-2.4%, -2.4%]	1

Cycles

Results (secondary -25.7%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.7%	[0.7%, 0.7%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-29.4%	[-67.4%, -0.6%]	7
All ❌✅ (primary)	-	-	0

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 495.518s -> 495.667s (0.03%)
Artifact size: 394.42 MiB -> 394.40 MiB (-0.00%)