`Vec::from_iter` specialization is slower on `TrustedLen` iterators from `thin-vec`

In https://github.com/rust-lang/rust/pull/159974, https://github.com/rust-lang/rust/pull/159975, we found out that updating and enabling "unstable" feature flag in thin-vec makes some new-solver benchmarks slower.

I tried to measure this in isolation on the old version in https://github.com/rust-lang/rust/pull/160010, where the old "unstable" flag only adds TrustedLen impl to thin_vec::IntoIter and thin_vec::Drain. The results for only adding the flag are here and they reproduce the regression: https://github.com/rust-lang/rust/pull/160010#issuecomment-5093128897

Based on the cachegrind diffs, the culprit is mostly in this iterator chain:

https://github.com/rust-lang/rust/blob/d3ea0356830dc019038110eba0bc8f790aaef2e4/compiler/rustc_trait_selection/src/solve/fulfill.rs#L185-L198

pending.drain() creates thin_vec::Drain iterator. The rest of it are iterators from std.

Adding no-op skip(0) before the collect() call disables the specialization (because Skip doesn't implement TrustedLen) and fixes most of the regression: https://github.com/rust-lang/rust/pull/160010#issuecomment-5098416135, but not all of it, so it looks like the problem is not specific to this chain.

If I understand the specialization resolution correctly, the problematic from_iter impl is this one:

https://github.com/rust-lang/rust/blob/d3ea0356830dc019038110eba0bc8f790aaef2e4/library/alloc/src/vec/spec_from_iter_nested.rs#L50-L62

Note that https://github.com/rust-lang/rust/pull/160005 optimizes some of the next-solver data structures, so it's possible the issue will no longer reproduce on this code on main as visibly, but the issue might still exist.

I currently don't have time to research this further and create smaller reproducer, but maybe somebody who has more experience with these specializations has a better idea for what's going on.

Start with the iterator chain in compiler/rustc_trait_selection/src/solve/fulfill.rs and the Vec specialization in library/alloc/src/vec/spec_from_iter_nested.rs. Reproduce the regression with the thin-vec unstable feature and inspect the cachegrind comparisons from PR #160010. Done means identifying the TrustedLen specialization cause and confirming the relevant benchmark no longer regresses.