Cardano Updates

Uses the `Seq` type instead of a `list`, which makes it a little faster,
on my computer, in some circumstances:

### Scenario 1

Before (`--number-of-txs 300`):

```
> cabal run bench-e2e -- datasets --number-of-txs 300 --output-directory /tmp/bench-before-300
Average confirmation time (ms): 17.457022511
P99: 29.295330019999994ms
P95: 22.807852549999993ms
P50: 16.398083ms
```
41s clock time as measured by my terminal.

After (`--number-of-txs 300`):

```
> cabal run bench-e2e -- datasets --number-of-txs 300 --output-directory /tmp/bench-after-300
Average confirmation time (ms): 17.389314338
P99: 27.240633599999995ms
P95: 22.911740449999996ms
P50: 16.474119ms
```

40s clock time.

_Conclusion_: Almost no difference.

### Scenario 2

Before (`--number-of-txs 3000`):

```
> cabal run bench-e2e -- datasets --number-of-txs 3000 --output-directory /tmp/bench-before-3000
Average confirmation time (ms): 40.479327468
P99: 57.356654989999996ms
P95: 52.25816309999999ms
P50: 40.995065ms
```

2min 37s clock time.

After (`--number-of-txs 3000`):

```
> cabal run bench-e2e -- datasets --number-of-txs 3000 --output-directory /tmp/bench-after-3000
Average confirmation time (ms): 17.389314338
P99: 27.240633599999995ms
P95: 22.911740449999996ms
P50: 16.474119ms
```

1min 25s clock time.

_Conclusion_: Roughly twice as fast!

-- Edit:

<img width="1200" height="900" alt="image"
src="https://github.com/user-attachments/assets/c8f88948-acee-4f90-9fa3-4f46515f6119"
/>

Signed-off-by: Sasha Bogicevic <[email protected]>

Signed-off-by: Sasha Bogicevic <[email protected]>

  Introduce HydraNodeConfig record type and smart constructors
  (mkSoloConfig, mkTwoPartyConfigs, mkThreePartyConfigs) to replace
  positional arguments in withHydraNode, withHydraNodeCatchingUp, and
  withUnsyncedHydraNode. Update all call sites in Scenarios, EndToEndSpec,
  ChainObserverSpec, TUISpec, and Bench/EndToEnd accordingly.

Signed-off-by: Sasha Bogicevic <[email protected]>

The `mithri`l `2617.0` release switched all nodes to static musl builds.
As a side effect, mithril-client-cli is no longer exposed as a
standalone Nix package for non-Linux platforms. It only exists under
`packages.x86_64-linux`. This broke nix develop on `aarch64-darwin`.


The one real downside of current solution is that the fallback pulls in
the entire mithril package (aggregator, signer, end-to-end tooling,
etc.) just to get `mithril-client`.

<!-- Describe your change here -->

---

<!-- Consider each and tick it off one way or the other -->
* [ ] CHANGELOG updated or not needed
* [ ] Documentation updated or not needed
* [ ] Haddocks updated or not needed
* [ ] No new TODOs introduced or explained herafter

---------

Co-authored-by: Noon <[email protected]>

Signed-off-by: Sasha Bogicevic <[email protected]>

  newLocalUTxO was carried on TransactionAppliedToLocalUTxO,
  SnapshotRequested, and DecommitRecorded but is fully derivable from the
  event's tx (or snapshot fields) plus current state. Aggregate now
  recomputes it via a new IsTx method `applyTxTo` (pure UTxO arithmetic,
  no ledger validation).

  Removes O(N²) on-disk growth on Growing-UTxO workloads. JSON format is
  forward/backward compatible — old event stores still parse.

  Measured (bench-e2e --utxo-size Growing -n 2000):
  - hydra.db:         1.6 GB → 526 MB  (-67%)
  - hydra-node-0.log: 2.1 GB → 1.1 GB  (-48%)
  - avg confirm:      92.9 ms → 84.1 ms (-9%)

Signed-off-by: Sasha Bogicevic <[email protected]>