fix(riscv): keep no-C control flow on the interpreter path

[carlosqwqqwq]

fix(riscv): keep no-C control flow on the interpreter path

Closes #217

Summary

Without C, IALIGN=32: JAL/branch targets must be 4-byte aligned and JALR targets are runtime-computed, so none of the three can be emitted as a translated branch without an alignment check. The lifter now returns Unsupported for these cases so the interpreter delivers the architectural instruction-address-misaligned trap.

Validation

• Regression covers misaligned JAL/branch targets and JALR under a no-C profile (must not lift), and the same words under C (must lift).
• cargo test --lib: 8214 passed, 0 failed.

Notes

Translated-path IALIGN gate, distinct from the no-C JALR handling in the direct interpreter.

[19h]

I reviewed the focused commit bec7348f against issue #217. The fail-closed behavior is sound: a statically misaligned no-C JAL/branch target falls back, and all no-C JALR instructions fall back because the target alignment is runtime-dependent; legal C-enabled controls remain liftable.

Two changes are needed before this is merge-ready:

1. This PR still contains 55043ddc, the broad refactor(riscv): split vector semantics out of cpu.rs #202 vector-module split that was closed as superseded by the focused structure in fix(riscv): enforce RVV reserved encoding constraints #203. Please rebase onto current master and drop that commit so this PR contains only the two SMIR lifter files.
2. The regression has two adjacent #[test] attributes. I ran the focused filter and confirmed Cargo registers and executes the same test twice (2/2 with the identical name). Please remove the duplicate marker.

Please also add aligned no-C JAL and conditional-branch controls so the regression proves the new fallback is limited to the misaligned static targets. This fix is non-RVV and should remain independent from #203.

[19h]

Thank you @carlosqwqqwq for identifying the no-C IALIGN gap and proposing a fix. We independently reimplemented and validated this behavior in #231, where it is consolidated with #220, #222, and #224. No commits from this branch were taken; the rollup was derived from the issue reproducer and the RISC-V ISA requirements. To keep review and CI on one maintained branch, I’m closing this PR as superseded by #231. You are credited as co-author on the consolidation commit—please review #231, particularly its conservative dynamic-JALR fallback.