test: parameterize Phase 2 closure tests cross-engine (PR4 of 5)

[danieljohnmorris]

Summary

Phase 2 PR4 of 5. After #384 (VM closure support), #385 (Cranelift closure parity), and #387 (HOF native closure dispatch) all merged, the existing inline-lambda regression suite still ran exclusively on the tree walker. This PR parameterises the suite over tree, VM, and Cranelift so VM and JIT regressions get caught at the same surface the tree walker does.

No production code changes. Test infrastructure only.

What's in the diff

Commit 1 - test: parameterize inline lambda regression suite over engines

• tests/regression_inline_lambda.rs: split run_ok into a run_engine primitive plus two callers, run_all (default, iterates tree + VM + Cranelift) and run_tree_only (escape hatch with required justification).
• 13 of the 21 tests flip directly to run_all (map / flt / fld and grouped-expression / no-regression coverage).
• 8 tests that exercise srt closure dispatch stay tree-only with a TODO PR3c follow-up comment pointing at feature: srt/grp/uniqby off tree-bridge (Phase 2 PR3c) #391 as the unblock. PR3c is in flight; once it lands the flip is a one-line swap.

Commit 2 - test: generalize closure capture coverage to all engines

• tests/coverage_vm_mod.rs: the two closure_capture_*_tree tests were tree-only relics from the pre-feature: VM closure support (Phase 2 PR1) #384 era. Generalised to ENGINES_ALL, dropped the _tree suffix.

Examples skip cleanup

Grep of examples/engine-skip: shows 12 hits. None of them are stale relative to this PR:

• closure-bind.ilo, uniqby.ilo, srt-by-key.ilo, grp-basics.ilo, sort-by-key.ilo - all gated on srt/grp/uniqby tree-bridge, waiting on feature: srt/grp/uniqby off tree-bridge (Phase 2 PR3c) #391.
• match-in-loop.ilo, list-ops.ilo, strings.ilo, data.ilo, builtins-as-hof.ilo - unrelated to closure capture.

No examples were edited in this PR. Once #391 merges, a follow-up sweep can drop the relevant skips.

Test plan

• cargo test --release --features cranelift --test regression_inline_lambda - 21 / 21 pass
• cargo test --release --features cranelift --test coverage_vm_mod - all pass including the 2 generalised closure tests
• cargo test --release --features cranelift full suite - all green
• cargo fmt --check
• cargo clippy --release --features cranelift --tests -- -D warnings

Follow-ups

• When feature: srt/grp/uniqby off tree-bridge (Phase 2 PR3c) #391 merges, swap the 8 run_tree_only calls in regression_inline_lambda.rs to run_all and drop the TODO comments.
• Same trigger lets closure-bind.ilo / srt-by-key.ilo / uniqby.ilo / grp-basics.ilo / sort-by-key.ilo drop their engine-skip lines.

[codecov]

❌ 4 Tests Failed:

Tests completed	Failed	Passed	Skipped
5148	4	5144	0

View the top 3 failed test(s) by shortest run time

ilo::regression_inline_lambda::closure_capture_single_var_filter

Stack Traces | 0.015s run time

thread 'closure_capture_single_var_filter' (38671) panicked at tests/regression_inline_lambda.rs:51:5:
ilo --run-cranelift failed for `f xs:L n thr:n>L n;flt (x:n>b;>x thr) xs`: stderr=error: unrecognised flag '--run-cranelift'. Use 'ilo --help' for valid flags. To pass it as a literal arg, separate with '--' first.

note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

ilo::regression_inline_lambda::closure_capture_multiple_vars

Stack Traces | 0.016s run time

thread 'closure_capture_multiple_vars' (38667) panicked at tests/regression_inline_lambda.rs:51:5:
ilo --run-cranelift failed for `f xs:L n lo:n hi:n>L n;flt (x:n>b;&(>=x lo) <=x hi) xs`: stderr=error: unrecognised flag '--run-cranelift'. Use 'ilo --help' for valid flags. To pass it as a literal arg, separate with '--' first.

note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

ilo::regression_inline_lambda::closure_capture_in_fld

Stack Traces | 0.018s run time

thread 'closure_capture_in_fld' (38654) panicked at tests/regression_inline_lambda.rs:51:5:
ilo --run-cranelift failed for `f xs:L n weight:n>n;fld (a:n x:n>n;+a *x weight) xs 0`: stderr=error: unrecognised flag '--run-cranelift'. Use 'ilo --help' for valid flags. To pass it as a literal arg, separate with '--' first.

note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

ilo::regression_inline_lambda::closure_capture_in_map

Stack Traces | 0.018s run time

thread 'closure_capture_in_map' (38661) panicked at tests/regression_inline_lambda.rs:51:5:
ilo --run-cranelift failed for `f xs:L n bump:n>L n;map (x:n>n;+x bump) xs`: stderr=error: unrecognised flag '--run-cranelift'. Use 'ilo --help' for valid flags. To pass it as a literal arg, separate with '--' first.

note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

To view more test analytics, go to the Test Analytics Dashboard
_{📋 Got 3 mins? Take this short survey to help us improve Test Analytics.}