feat: Rank by Surplus Tracker

[sunce86]

Problem

Issue tracker for rank by surplus implementation.
Solutions ranking needs to be changed from rank by score to rank by surplus.

Here is how surplus is defined in the driver:
surplus- raw surplus that is defined by original clearing prices provided by solvers
surplus' = surplus - surplus_fee
surplus'' = surplus' - protocol_fee

So, we want to rank by surplus'.

Tasks:

• Implement ranking by surplus in the driver - ON HOLD: Calculate score as per "rank by surplus" CIP #2406
• Add configuration parameter in the driver, that will be used to switch to CIP38 rules.

Nice to have:

• [After release] Drop score wherever used (from solvers interface for example). Keep it in the driver interface since it will be reused to report surplus to autopilot since autopilot won't calculate this but it will trust the driver - chore: Cleanup driver score function #2493
• Also double check the delivered surplus in the autopilot based on the onchain data (to verify that the promised solution was delivered) - feat: Implement rank by surplus calculation in autopilot #2494
• [After release] Remove temporary configuration - chore: Remove temporary configuration for CIP38 #2462

[fleupold]

Change the way the autopilot ranks the solutions

Currently the autopilot just accepts the score from the driver as true. Do we want to change this and have the autopilot compute surplus and fees?

[sunce86]

Change the way the autopilot ranks the solutions

Currently the autopilot just accepts the score from the driver as true. Do we want to change this and have the autopilot compute surplus and fees?

I would prefer so. If autopilot can calculate the surplus (or any other value we use for ranking) from the solution itself, then there is no need for additional field as part of the /solve response.

[sunce86]

This is somewhat aligned with the idea explained here: #2225

[fleupold]

there is no need for additional field as part of the /solve response.

Well, instead (at least as outlined in #2225) there is need for the entire encoded settlement (which is much more than the score). I think if the value can be computed by what we already send then sure let the autopilot verify it, but if not, I think it's fine to trust the driver with this (at the end of the day the driver could also "lie" about their encoded settlement)

[sunce86]

All tasks covered with individual issues so closing this one.