new block cache and much more...

- more reliable code invalidation detection
- blocks aren't stopped at any branch, but are being followed
if possible to get larger blocks
- idle loop recognition
- optimised literal loads, load/store cycle counting
 and loads/stores from constant addresses

src/ARM_InstrInfo.h

@@ -226,18 +226,27 @@ enum
     flag_V = 1 << 0,
 };
 
+enum
+{
+    special_NotSpecialAtAll = 0,
+    special_WriteMem,
+    special_WaitForInterrupt
+};
+
 struct Info
 {
     u16 DstRegs, SrcRegs;
     u16 Kind;
 
+    u8 SpecialKind;
+
     u8 ReadFlags;
     // lower 4 bits - set always
     // upper 4 bits - might set flag
     u8 WriteFlags;
 
     bool EndBlock;
-    bool Branches()
+    bool Branches() const
     {
         return DstRegs & (1 << 15);
     }