Avoid clearing sequential bits in a loop

[vkrasnov]

Fixes #54

[Lekensteyn]

See inline comment for one issue.

Another missing thing to fully address #54 is an explicit check against reaching message limits. That would also remove any concern for integer overflows.

[Lekensteyn]

LGTM. Would you consider adding a check for counters past the limits in #54 though? As you said it won't happen in practice, but since it is not permitted by the spec it should not hurt to block it either.

[vkrasnov]

I think the implementation (or did I see it somewhere else?) says that the numbers are a placeholder in case a realistic limit is needed. But since those are hardcoded, the value can only change with the implementation anyway

[Lekensteyn]

At least the Linux kernel implementation does have an explicit check for those insanely large numbers. In the protocol spec (whitepaper), the counter starts at 0 and increases for every data packet. So it seems reasonable to me to check the counter against the REJECT-AFTER-MESSAGES limit.