Improve malicious security implementation for M2A/A2M conversion protocols

[th4s]

The current replay functionality added by tlsnotary/tlsn#131 allows to detect malicious behavior of the sender.

However, we need a thorough analysis if the current implementation is sufficient. This involves to determine what security checks we need and why. Especially it is currently not clear if we need to

• seed the rng by a cointoss
• verify ot envelopes

Once we figure this out, we need to add possible new checks to the current implementation.