[x64] Fix AVX-512 VECTOR_DENORMFLUSH merge-masking & FpRangeSign values

Copy src1 into dest before the masked vrangeps so that unmasked lanes
(normal/inf/NaN values) are preserved via merge-masking. Previously,
when the register allocator assigned different physical registers for
dest and src1, unmasked lanes would retain stale values from dest.

Tested with Intel SDE and AVX-512 cpu to verify.

Also shift unused FpRangeSign constants into bits [3:2] of the vrangeps
immediate, matching the Intel encoding. Only one used in the code was 0
so it was already correct by accident but using the others in the future
would produce incorrect results.

src/xenia/cpu/testing/vector_flush_denorm_test.cc

@@ -43,3 +43,44 @@ TEST_CASE("VECTOR_DENORMFLUSH_F32", "[instr]") {
              REQUIRE(result == vec128f(-FLT_MIN));
            });
 }
+
+// Test VECTOR_DENORMFLUSH with register pressure: input is kept
+// alive past the denormflush, forcing dest and src into different
+// physical registers. Verifies correct passthrough of normal values
+// and preservation of the original input.
+TEST_CASE("VECTOR_DENORMFLUSH_F32_REGPRESSURE", "[instr]") {
+  TestFunction test([](HIRBuilder& b) {
+    auto input = LoadVR(b, 4);
+    StoreVR(b, 3, b.VectorDenormFlush(input));
+    StoreVR(b, 5, input);  // keep input alive past denormflush
+    b.Return();
+  });
+  // Mix of denormal and normal elements:
+  //   [0] 0x007FFFFF = positive denormal  -> should flush to +0.0
+  //   [1] 0x3F800000 = 1.0f (normal)      -> should pass through
+  //   [2] 0x807FFFFF = negative denormal  -> should flush to -0.0
+  //   [3] 0x40000000 = 2.0f (normal)      -> should pass through
+  test.Run(
+      [](PPCContext* ctx) {
+        ctx->v[4] = vec128i(0x007FFFFF, 0x3F800000, 0x807FFFFF, 0x40000000);
+      },
+      [](PPCContext* ctx) {
+        auto result = ctx->v[3];
+        REQUIRE(result ==
+                vec128i(0x00000000, 0x3F800000, 0x80000000, 0x40000000));
+        // Original input must also be preserved
+        REQUIRE(ctx->v[5] ==
+                vec128i(0x007FFFFF, 0x3F800000, 0x807FFFFF, 0x40000000));
+      });
+  // All normal values - with the bug, ALL lanes would be corrupted
+  // since k1 would be all-zero and merge-masking preserves dest, not src1
+  test.Run(
+      [](PPCContext* ctx) {
+        ctx->v[4] = vec128i(0x3F800000, 0x40000000, 0x40400000, 0x40800000);
+      },
+      [](PPCContext* ctx) {
+        auto result = ctx->v[3];
+        REQUIRE(result ==
+                vec128i(0x3F800000, 0x40000000, 0x40400000, 0x40800000));
+      });
+}