ffi 0.4: focused FP-aggregate (HFA) baseline — FQuad + DQuad

91/91 regression tests pass (+ffi-04-fp-struct).

Single-file regression net for the all-float / all-double aggregate
ABI path:

  FQuad — 16 B, 4×f32   (same slot as ffi-02's Vec4f)
  DQuad — 32 B, 4×f64   (UIEdgeInsets-shape — the f32-vs-f64 landmine)

Already nominally covered by ffi-02's Vec4f, but pinning it as a
focused single-file test means a future ABI rule change that breaks
the HFA path fails *this* test directly without a noisy drag-in from
the multi-shape baseline.

DQuad at 32 B straddles the AAPCS64 HFA limit (≤4 floats of same
type, total ≤64 B); it stays as a struct value passed through
v0..v3 rather than going indirect. The snapshot confirms the values
arrive intact.

vendors/ffi_fp_struct/ffi_fp_struct.c

@@ -0,0 +1,29 @@
+#include "ffi_fp_struct.h"
+
+FQuad ffi_fquad_make(float a, float b, float c, float d) {
+    FQuad r = { a, b, c, d };
+    return r;
+}
+
+FQuad ffi_fquad_reverse(FQuad v) {
+    FQuad r = { v.d, v.c, v.b, v.a };
+    return r;
+}
+
+float ffi_fquad_sum(FQuad v) {
+    return v.a + v.b + v.c + v.d;
+}
+
+DQuad ffi_dquad_make(double a, double b, double c, double d) {
+    DQuad r = { a, b, c, d };
+    return r;
+}
+
+DQuad ffi_dquad_reverse(DQuad v) {
+    DQuad r = { v.d, v.c, v.b, v.a };
+    return r;
+}
+
+double ffi_dquad_sum(DQuad v) {
+    return v.a + v.b + v.c + v.d;
+}

vendors/ffi_fp_struct/ffi_fp_struct.h

@@ -0,0 +1,20 @@
+// Focused FP-aggregate (HFA) FFI baselines. Distinct from the int-aggregate
+// register-coercion paths because all-float / all-double structs of ≤4 fields
+// stay as struct values in LLVM and are passed/returned via the float
+// register file (AAPCS64 v0..v3; SysV AMD64 xmm0..xmm7). This was the
+// `UIEdgeInsets`-as-f32-vs-f64 landmine — pinned here so a future ABI rule
+// change that wrecks the FP path fails this test directly.
+//
+//   FQuad   — 16 B, four float    (small HFA; same slot as Vec4f)
+//   DQuad   — 32 B, four double   (UIEdgeInsets-shape HFA)
+
+typedef struct { float a; float b; float c; float d; }              FQuad;
+typedef struct { double a; double b; double c; double d; }          DQuad;
+
+FQuad  ffi_fquad_make   (float a, float b, float c, float d);
+FQuad  ffi_fquad_reverse(FQuad v);
+float  ffi_fquad_sum    (FQuad v);
+
+DQuad  ffi_dquad_make   (double a, double b, double c, double d);
+DQuad  ffi_dquad_reverse(DQuad v);
+double ffi_dquad_sum    (DQuad v);