Implement Smulhi for interpreter

Implemented `Smulhi` for the Cranelift interpreter, performing signed
integer multiplication and producing the high half of a double-length
result.

Copyright (c) 2021, Arm Limited

cranelift/filetests/filetests/runtests/smulhi.clif

@@ -0,0 +1,32 @@
+test interpret
+test run
+target aarch64
+set enable_simd
+target x86_64 machinst
+
+function %smulhi_i16(i16, i16) -> i16 {
+block0(v0: i16, v1: i16):
+    v2 = smulhi v0, v1
+    return v2
+}
+; run: %smulhi_i16(-2, -4) == 0
+; run: %smulhi_i16(2, -4) == -1
+; run: %smulhi_i16(32767, 32767) == 16383
+
+function %smulhi_i32(i32, i32) -> i32 {
+block0(v0: i32, v1: i32):
+    v2 = smulhi v0, v1
+    return v2
+}
+; run: %smulhi_i32(-500, -700) == 0
+; run: %smulhi_i32(500, -700) == -1
+; run: %smulhi_i32(2147483647, 2147483647) == 1073741823
+
+function %smulhi_i64(i64, i64) -> i64 {
+block0(v0: i64, v1: i64):
+    v2 = smulhi v0, v1
+    return v2
+}
+; run: %smulhi_i64(-4294967295, -4294967295) == 0
+; run: %smulhi_i64(4294967295, -4294967295) == -1
+; run: %smulhi_i64(9223372036854775807, 9223372036854775807) == 4611686018427387903