perf: avoid discarded simple list assignment results

[fglock]

Summary

Avoid materializing the discarded RuntimeArray result for the common simple-scalar list assignment path when the compiler knows the assignment is in void context. The observable-result path remains unchanged.

This is deliberately a general argument-frame/result-allocation reduction, rather than a specialization for a particular method implementation.

Validation

• New focused regression test passed on system Perl: /tmp/perl-void-list-assignment-20260919.log.
• Full project make passed for the exact measured source commit: /tmp/make-perf-void-list-assignment-measure-20260919.log.
• make check-links passed after the changelog entry: /tmp/check-links-perf-void-list-assignment-20260919.log.
• Normal production-method screen used stable checksums and showed a 1.115x geometric-mean improvement versus adjacent baseline brackets.
• JFR allocation sampling on that workload reduced sampled RuntimeArray allocations from 318 to 117.
• A low-priority, production-condition full portfolio was protocol-compliant and stable. Against the clean parent portfolio, Closure improved 0.574 to 0.706x Perl, Method 0.152 to 0.157x, Numeric 0.265 to 0.302x, String 0.328 to 0.372x, Life 0.382 to 0.484x, and JSON 0.031 to 0.034x; Regex was effectively flat/slightly lower (0.511 to 0.505x). The portfolio remains below the project-wide 1.05x-Perl acceptance target, so this is qualified evidence rather than an acceptance claim.

Stack

This PR is intentionally based on #1439, and therefore includes the preceding unsigned-shift and foreach-alias work from #1438/#1439. Review the final commit independently; merge after its base PR is merged.