Assertion failure in long-range HFC ticking

Consider ticking an `ExtLedgerState` of a `HardForkBlock` across an era/epoch boundary where the tip slot is more than one safe zone before the era/epoch boundary. This is currently (current `main`, ie 0dc2699a6907c0a95371270e685b6175dbd30e82) triggering an assertion failure (concretely, ticking from `133531000` to `133660800` across the Conway HF):

```
CallStack (from HasCallStack):
error, called at src/ouroboros-consensus/Ouroboros/Consensus/Util/Assert.hs:11:30 in ouroboros-consensus-0.20.1.0-inplace:Ouroboros.Consensus.Util.Assert
assertWithMsg, called at src/ouroboros-consensus/Ouroboros/Consensus/HardFork/History/Util.hs:32:38 in ouroboros-consensus-0.20.1.0-inplace:Ouroboros.Consensus.HardFork.History.Util
countSlots, called at src/ouroboros-consensus/Ouroboros/Consensus/HardFork/History/Summary.hs:130:25 in ouroboros-consensus-0.20.1.0-inplace:Ouroboros.Consensus.HardFork.History.Summary
slotToEpochBound, called at src/ouroboros-consensus/Ouroboros/Consensus/HardFork/Combinator/State/Infra.hs:254:15 in ouroboros-consensus-0.20.1.0-inplace:Ouroboros.Consensus.HardFork.Combinator.State.Infra
applySafeZone, called at src/ouroboros-consensus/Ouroboros/Consensus/HardFork/Combinator/State/Infra.hs:222:29 in ouroboros-consensus-0.20.1.0-inplace:Ouroboros.Consensus.HardFork.Combinator.State.Infra
reconstructSummary, called at src/ouroboros-consensus/Ouroboros/Consensus/HardFork/Combinator/State.hs:167:7 in ouroboros-consensus-0.20.1.0-inplace:Ouroboros.Consensus.HardFork.Combinator.State
epochInfoPrecomputedTransitionInfo, called at src/ouroboros-consensus/Ouroboros/Consensus/HardFork/Combinator/Protocol.hs:192:12 in ouroboros-consensus-0.20.1.0-inplace:Ouroboros.Consensus.HardFork.Combinator.Protocol
```
(I added some `HasCallStack` constraints for this)

Note that this happens as part of ticking to tick the protocol state, after the ledger state has already been ticked.

We should think more about such "long-range" HFC ticking and properly document whether we do (not) expect it to work, and point this out in the code.

---

This luckily isn't relevent for the current node as we never perform such long-range ticks: Both sources of new blocks (ChainSync/BlockFetch and forging) first require *forecasting* to the target slot, which only succeeds within a safe zone, so any subsequent tick won't be a long-range tick.

Start with the long-range ticking path in Ouroboros.Consensus.HardFork.Combinator.State.Infra, especially applySafeZone and slotToEpochBound, then read reconstructSummary in State.hs and epochInfoPrecomputedTransitionInfo in Protocol.hs. Reproduce the tick from 133531000 to 133660800 and document whether long-range HFC ticking is supported, including the relevant code locations.

Written by the indexing model from the issue text.