Fixed bound on memcpy loop in thumb,arm,aarch64

[themaks]

Hello,

There was a +10 years-old out-of-bound error in the memcpy implementation of shellcraft for thumb, arm and aarch64: the loop always copied one byte more than instructed in n.

By the way, there also is a problem when we supply an n in [2**31;2**32-1] (since the comparison is signed), and also when n==0 (one byte is copied nonetheless), but I guess we can leave those bugs for the sake of shellcode's simplicity 🙃

Cheers

[RocketMaDev]

Since n==0, could we return an empty string? As for "n in [2**31;2**32-1]", I guess noboby will copy that much bytes. And you could update the CHANGELOG if you would like.

[peace-maker]

Thank you! Nice catch. Can you rebase and target the stable branch instead please? Let me know if you need assistance with git.