Optimize issuing advance engine fcU for prepare payload

[g11tech]

Currently the engine fcU is issued just after a successful block import (#3965), predicting the new head to be the head in the next slot if BN has any beacon proposer. Normally this would work just fine, but this isn't sufficient and the following scenarios needs to be addressed:

• wait till the last interval of slot before issuing fcU unless the head is already last imported
• issue the fcU
  • for skipped slot
  • for head change

[twoeths]

Right now PrecomputeNextEpochTransitionScheduler does not work after the merge due to that. I suppose we should make this work as a separate scheduler like in lighthouse and do precompute next epoch transition in this scheduler too https://github.com/sigp/lighthouse/blob/564d7da656803f5e06e53a303972580be54500bf/beacon_node/beacon_chain/src/proposer_prep_service.rs#L18

[g11tech]

Right now PrecomputeNextEpochTransitionScheduler does not work after the merge due to that. I suppose we should make this work as a separate scheduler like in lighthouse and do precompute next epoch transition in this scheduler too https://github.com/sigp/lighthouse/blob/564d7da656803f5e06e53a303972580be54500bf/beacon_node/beacon_chain/src/proposer_prep_service.rs#L18

my idea is not NOT run PrecomputeNextEpochTransitionScheduler but run PrecomputeNextSlotTransitionScheduler and do an fcU there if not epoch transition else first do epoch transition and then chain fcU to it. what do you think @tuyennhv ?

or we do slot processing on a copy ?

[g11tech]

if you want to take this up @tuyennhv feel free!

[twoeths]

my idea is not NOT run PrecomputeNextEpochTransitionScheduler but run PrecomputeNextSlotTransitionScheduler and do an fcU there if not epoch transition else first do epoch transition and then chain fcU to it. what do you think @tuyennhv ?

yeah we're on the same page, we only need 1 scheduler, yeah I can take this @g11tech