x64: Lower extractlane, scalar_to_vector, and splat in ISLE (#4780)

Lower extractlane, scalar_to_vector and splat in ISLE.

This PR also makes some changes to the SinkableLoad api
* change the return type of sink_load to RegMem as there are more functions available for dealing with RegMem
* add reg_mem_to_reg_mem_imm and register it as an automatic conversion

cranelift/codegen/src/isa/x64/inst/emit_tests.rs

@@ -78,6 +78,17 @@ impl Inst {
             dst: WritableXmm::from_writable_reg(dst).unwrap(),
         }
     }
+
+    // TODO Can be replaced by `Inst::move` (high-level) and `Inst::unary_rm_r` (low-level)
+    fn xmm_mov(op: SseOpcode, src: RegMem, dst: Writable<Reg>) -> Inst {
+        src.assert_regclass_is(RegClass::Float);
+        debug_assert!(dst.to_reg().class() == RegClass::Float);
+        Inst::XmmUnaryRmR {
+            op,
+            src: XmmMem::new(src).unwrap(),
+            dst: WritableXmm::from_writable_reg(dst).unwrap(),
+        }
+    }
 }
 
 #[test]