x64: allow vector types in select move

As reported in #3173, the `select` instruction fails an assertion when it is given `v128` types as operands. This change relaxes the assertion to allow the same type of XMM move that occurs for the f32 and f64 types. This fixes #3173 in the old `lower.rs` code temporarily until the relatively complex `select` lowering can be ported to ISLE.

tests/misc_testsuite/simd/issue_3173_select_v128.wast

@@ -0,0 +1,10 @@
+(; See issue https://github.com/bytecodealliance/wasmtime/issues/3173. ;)
+
+(module
+  (func (export "select_v128") (result v128)
+    v128.const i32x4 0x00000000 0x00000000 0x00000000 0x00000000
+    v128.const i32x4 0x00000000 0x00000000 0x00000000 0x00000000
+    i32.const 0
+    select))
+
+(assert_return (invoke "select_v128") (v128.const i32x4 0 0 0 0))