Reconsider HFC requirement that ticking cannot make a transition become known

[amesgen]

Right now, the HFC makes use of the fact that ticking can't let a transition become known:

ouroboros-consensus/ouroboros-consensus/src/ouroboros-consensus/Ouroboros/Consensus/HardFork/Combinator/Ledger.hs

Lines 130 to 131 in 106ce89

	-- o 'TransitionUnknown'. If the transition is unknown, then it
	-- cannot become known due to ticking. In this case, we record

Right now, this is true due to "block counting" (see eg ShelleyTransition). On the other hand, alternatives to block counting (see #389) likely require relaxing this assumption.

For more context, see:

• This HWW section ("The logic used in this function must be invariant with respect to ticking. [...]").
• The dimension "DST" in the HWW section added in docs: draft HardForkCombinatorRollbackMonotonicityProposal.md #346
• Stop assuming that ticking cannot influence the TransitionInfo #365 for a draft implementation of this idea. At a high level, singleEraTransition now takes a slot to which the ledger state will be ticked next.