Actually add RCX, plus RDI and RSI on Unix.

And switch to a register order that consistently prefers callee-save to
caller-save.  phire suggested putting rdi/rsi first, even though they're
caller-save, to save code space; this is more conservative and I can do
that later.

Source/Core/Core/PowerPC/Jit64/JitRegCache.cpp

@@ -200,9 +200,9 @@ const int* GPRRegCache::GetAllocationOrder(size_t& count)
 	{
 		// R12, when used as base register, for example in a LEA, can generate bad code! Need to look into this.
 #ifdef _WIN32
-		RSI, RDI, R13, R14, R8, R9, R10, R11, R12, //, RCX
+		RSI, RDI, R13, R14, R8, R9, R10, R11, R12, RCX
 #else
-		R13, R14, R8, R9, R10, R11, R12, //, RCX
+		R12, R13, R14, RSI, RDI, R8, R9, R10, R11, RCX
 #endif
 	};
 	count = sizeof(allocationOrder) / sizeof(const int);