Refactor delta proof generation code

[murisi]

@heueristik brought up the following points while reviewing #235 :

1. Reduce the number of fuzzed variables. Constrain them only at the beginning of tests and don’t reset/replace/overwrite them in deeper levels of the test.
2. Split delta generation code from tests into a separate contract (and later into a library containing only internal pure functions). The delta generation library code can check inputs and throw custom errors but can not contain fuzzing variable constraints.
3. Remove the dependency of the delta proof generation code on vm (tests can and should still use vm)
   1. Move vm.assume(...) and bound(...) statements to the beginning of the fuzzing tests and deduplicate them. If they repeat, encapsulate them into functions with a meaningful names.
   2. Try to use elliptic curve library functions instead of vm.createWallet or vm.sign. Note that we can bring more functions from import {EllipticCurve} from "@elliptic-curve-solidity/contracts/EllipticCurve.sol"; into library EllipticCurveK256 if this is needed.
4. Encapsulate repeated test code into functions.
5. Rework the instance struct and replace int256 quantity by uint128 quantity and bool consumed
   • This should also simplify the transformations that need to be done

[murisi]

Implementing point (3) and (3ii) caused a slowdown in the test suite from 16.13s to 88.32s. (To see this, check out 603048d , replace all genProofs with DeltaGen.genProofs and genInstances with DeltaGen.genInstance.) This is probably due to the fact elliptic-curve-solidity is implemented in Solidity in contrast to Foundry cheatcodes which seem to be implemented in Rust.

Points (3) and (3ii) have now been superseded by recommendation to parameterize the delta generation functions with Vm vm parameter so that secp256k1 functions can be called through that. This recommendation has been implemented in 3397cbd .