CI: cargo-test red on main since #4957 — typed_shape_descriptors::bounded_integer_array_store_omits_layout_note_and_barrier

[proggeramlug]

main has a failing required check: cargo test -p perry-codegen --test typed_shape_descriptors fails with

thread 'bounded_integer_array_store_omits_layout_note_and_barrier' panicked at crates/perry-codegen/tests/typed_shape_descriptors.rs:430:
bounded numeric array store should route through the raw-f64 payload helper

Bisection

• 3fbeea978 (fix(hir): self-named native member base, fix(hir): self-named native member base no longer OOM-loops codegen (#4908) #4955) — passes (15/15)
• 90bc7bc68 (Optimize numeric array raw payload helpers, Optimize numeric array raw payload helpers #4957) — fails (14/15)

So #4957 changed the raw-f64 payload-helper routing for bounded numeric array stores without updating (or while genuinely regressing) the descriptor expectation pinned by this test. Every PR branch since inherits the red cargo-test check from the merge with main — e.g. #4979's cargo-test failed solely on this (all of its own suites green; verified the failure reproduces identically at its merge-base and on origin/main HEAD, Linux CI and macOS local).

Either the optimization needs to keep routing bounded stores through the raw-f64 payload helper, or the test's expectation should be updated to the new intended IR shape — whichever #4957 meant.

[proggeramlug]

Duplicate of #4985, and verified fixed on current main — typed_shape_descriptors::bounded_integer_array_store_omits_layout_note_and_barrier passes (1/1) on a pristine checkout. See #4985 for details.