x86 sse3 simde_mm_loaddup_pd: avoid UB in Neon A32V7 impl

Author: mr-c

simde/x86/sse3.h

@@ -434,7 +434,7 @@ simde_mm_loaddup_pd (simde_float64 const* mem_addr) {
     #if defined(SIMDE_ARM_NEON_A64V8_NATIVE)
       r_.neon_f64 = vdupq_n_f64(*mem_addr);
     #elif defined(SIMDE_ARM_NEON_A32V7_NATIVE)
-      r_.neon_i64 = vdupq_n_s64(*HEDLEY_REINTERPRET_CAST(int64_t const*, mem_addr));
+      r_.neon_i64 = vdupq_n_s64(HEDLEY_REINTERPRET_CAST(int64_t, *mem_addr));
     #elif defined(SIMDE_LOONGARCH_LSX_NATIVE)
       r_.lsx_i64 = __lsx_vldrepl_d(mem_addr, 0);
     #else