Fuzz testing the protocol parser

[mhanne]

See #75 - this actually happened to me by accident (I was feeding it a block with the magic bytes still in front).

Maybe we should add some fuzz-testing for the protocol parser that just feeds it (semi?) random data and makes sure it doesn't fail in unexpected ways.

Any thoughts?

I think I have heard about such a thing for bitcoind once, but I can't find it anymore..