implement float -> unsigned casts

amd64 lacks instruction for this so it has to be implemented with
float -> signed casts. The approach is borrowed from llvm.

arm64/emit.c

@@ -89,7 +89,9 @@ static struct {
 	{ Ocast,   Ks, "fmov %=, %W0" },
 	{ Ocast,   Kd, "fmov %=, %L0" },
 	{ Ostosi,  Ka, "fcvtzs %=, %S0" },
+	{ Ostoui,  Ka, "fcvtzu %=, %S0" },
 	{ Odtosi,  Ka, "fcvtzs %=, %D0" },
+	{ Odtoui,  Ka, "fcvtzu %=, %D0" },
 	{ Oswtof,  Ka, "scvtf %=, %W0" },
 	{ Ouwtof,  Ka, "ucvtf %=, %W0" },
 	{ Osltof,  Ka, "scvtf %=, %L0" },