riscv: use exact rounding for tiny directed FMA results

[carlosqwqqwq]

riscv: use exact rounding for tiny directed FMA results

Fixes #308

Commit message

riscv: use exact rounding for tiny directed FMA results

Description

The directed-rounding FMA path can double-round a subnormal product and
its error term, producing a wrong result and false UF. Use one 128-bit exact
rounding path for the affected fmadd.d range under RTZ/RDN/RUP, and restore
NX/UF from the exact result. The change is limited to src/util/fpu_lib.h;
validation artifacts stay out of the source diff. This patch applies independently
to the staging branch and does not depend on an unmerged FMA change.

Validation

• The witness runs on native RISC-V hardware, QEMU, and the affected RVVM build; the patched build matches both references.
• Both the interpreter and JIT lanes were exercised, and the patched build is limited to this root.

The patch is independently applicable to the staging branch. RNE/RMM tiny
results remain outside this patch's scope.