Correct handling for Trap #2 in SunOS executables, now that we know it is supposed to flush the cache. (Was thought to be "some obscure FPU operation".)