mpt: Unnecessary magic string in proofs

[rozbb]

Currently all MPT inclusion and non-inclusion proofs start with the string "mptproof". Why is this necessary? CT proofs don't have a magic string, and I don't know of any issues that people have run into about their non-self-describing-ness.

Concretely, the 8B are not likely to make a huge difference (though, at scale, maybe you could make that argument). But it just seems not necessary, so I figure I'd flag it anyway.

[warpfork]

My 2c would be a vote to keep this.

Where we can get self-description that helps user familiarity and debugging, and without significant downsides, I'm in favor of it. This seems like a low-cost, edge-ward place to do so.

Since it's not included in a hash, I'm thinking it's a choice that doesn't seem to lock anything in. E.g. I reckon that if some larger protocol involves transferring so many of these that the extra 8 bytes is becoming inconvenient, that protocol can just... spec that it's not doing them anymore, and things should still generally be fine.