chore: Simulate settlements with current block timestamp + 12s

[fleupold]

Background

We sometimes see solution candidates that simulate well on the last slot of the current block and thus win the competition, but immediately fail in the first slot of the next block and therefore become invalid and never settle.

This can cause settlement delays if a solver keeps submitting solutions that only work on current but never on next.

Details

Examples for this are pool math involving block.timestamp, e.g.

Here is an example simulation

1. passing in last slot of current (phalcon)
2. failing in first slot of next (phalcon)

The only thing that changes between the two is block.timestamp. We should therefore use the timestamp of the next block when simulating solution candidates.

Acceptance criteria

Solutions that won't simulate successfully at the top of the next block get filtered out in the driver.