Domain Settlement Encoding

[fleupold]

Description

Main part of #2215. This PR ports the settlement encoding logic which is currently contained in the legacy solver crate's SettlemenEncoder to our domain.

It is importantly still missing application of slippage tolerance (which is contained in SlippageContext, porting which would have made this PR even larger.

Changes

• Introduce encoding module which exposes a function to take a solution and returns a settlement tx object
• Encode clearing prices, trades & interactions the way the settlement contract expects (guided by the boundary implementation)
• Implement revertable logic on Solutions (this also fixes feat: Always classify settlements containing ERC1271 orders as unsafe #2640)
• Make allowance computation internalisation aware (currently allowances for internalised interactions are added to the execution plan even if we are building an internalised one)
• Unify swap implementation of the liquidity types to always return Result (one returned Option)

How to test

• Change default encoding::Strategy from Boundary to Domain and run tests.

Important

We still need to implement the slippage computation to make this strategy fully working

Fixes #2215 & #2640

[squadgazzz]

LGTM

[MartinquaXD]

Code looks correct. Just a few nits.

[sunce86]

Adding a few unit tests where we compare the encoded settlement boundary vs domain would be very nice to prove the new code is indeed a refactor. Example solutions can be taken from prod logs.

[fleupold]

Adding a few unit tests where we compare the encoded settlement boundary vs domain would be very nice to prove the new code is indeed a refactor.

I think the testing of this feature can be seen by the e2e & driver integration tests passing (any handful of solution is also not going to miss subtle differences). What might be a good idea though is to "shadow encode" every boundary settlement with the domain encoder in prod for a while and log if there is any mismatch.