Fix zkapp

[L-as]

I redid the work in #78 because it changes the sequencer and has merge conflicts and some parts of it should have been separate.

[L-as]

Fix zkapp -> how is it fixing zkapp?

There were bugs in both the proof generation and circuit logic.

Also what is the current state on the fee payer? We are making bunch of changes to the core mechanism and I haven't seen it documented nor tested properly.

Check #78

I'm not sure what you mean by "core mechanism", but it has been documented in the code.

(* ZEKO NOTE: When we create a fee payer from a dummy fee payer that
   has the empty public key, we ignore the existing logic and replace it with
   a dummy account update with public key set to point near 123456789 *)

(* ZEKO NOTE: We don't mandate that the nonce is increased.
   This is useful for simplifying the L1's mempool, not relevant for us. *)

(* ZEKO NOTE: We don't require a signature for the fee payer *)

[MartinOndejka]

(* ZEKO NOTE: When we create a fee payer from a dummy fee payer that
   has the empty public key, we ignore the existing logic and replace it with
   a dummy account update with public key set to point near 123456789 *)

(* ZEKO NOTE: We don't mandate that the nonce is increased.
   This is useful for simplifying the L1's mempool, not relevant for us. *)

(* ZEKO NOTE: We don't require a signature for the fee payer *)

Documenting what exactly changed and how is user supposed to pay fees. This is significant change because it requires the tooling to adapt.

[L-as]

Documenting what exactly changed and how is user supposed to pay fees. This is significant change because it requires the tooling to adapt.

The user? This is for the sequencer. The sequencer needs to post a zkapp command without paying fees, and this is the solution to that.