Cardano Updates

Signed-off-by: Hyperledger Bot <[email protected]>

Signed-off-by: Hyperledger Bot <[email protected]>

Document all error stubs introduced for DijkstraEra and files
using deprecated Cardano.Api.Certificate that need migration.

Add Dijkstra era handling across the read library: block/tx
parsing, era value handling, tx generators, and output types.

Add Dijkstra era pattern matches and type adaptations across
the primitive ledger read layer, including certificates, mint,
outputs, scripts, and witness counting.

Update CHaP index-state, flake inputs, and align all package
dependency bounds with cardano-node 10.6.2 ecosystem.

Update api, balance-tx, network-layer, wallet, local-cluster,
integration, and unit test packages for compatibility with the
new cardano-node 10.6.2 APIs and DijkstraEra.

Document all error stubs introduced for DijkstraEra and files
using deprecated Cardano.Api.Certificate that need migration.

Update api, balance-tx, network-layer, wallet, local-cluster,
integration, and unit test packages for compatibility with the
new cardano-node 10.6.2 APIs and DijkstraEra.

Add Dijkstra era pattern matches and type adaptations across
the primitive ledger read layer, including certificates, mint,
outputs, scripts, and witness counting.

Add Dijkstra era handling across the read library: block/tx
parsing, era value handling, tx generators, and output types.

Update CHaP index-state, flake inputs, and align all package
dependency bounds with cardano-node 10.6.2 ecosystem.

## Summary

- Add `if: always()` cleanup step to the E2E workflow to remove
`/tmp/e2e-*` and `/tmp/test-cluster*` after each run

## Problem

When E2E tests get cancelled (timeout or `cancel-in-progress`), the
process is SIGKILL'd and `withSystemTempDirectory` cleanup never runs.
Each cancelled run leaks ~17GB in `/tmp`. Four stale dirs accumulated to
~57GB, filled the runner's 63GB tmpfs, and broke all CI jobs with
`database or disk is full` errors on the nix eval cache.

## Test plan

- [x] Manually cleaned `/tmp` on `builder-new` (93% → 1%)
- [x] Re-triggered failed CI runs
- [ ] Verify next E2E cancellation leaves no stale dirs

When E2E tests are cancelled (timeout or cancel-in-progress), the
process gets SIGKILL'd and withSystemTempDirectory cleanup never runs,
leaking ~17GB per run in /tmp/e2e-* and /tmp/test-cluster-* dirs.
This filled the runner's tmpfs and broke all CI jobs.

This is done for convenience of partial application with `mapWithKey`
like functions

Document all error stubs introduced for DijkstraEra and files
using deprecated Cardano.Api.Certificate that need migration.

Update api, balance-tx, network-layer, wallet, local-cluster,
integration, and unit test packages for compatibility with the
new cardano-node 10.6.2 APIs and DijkstraEra.

Add Dijkstra era pattern matches and type adaptations across
the primitive ledger read layer, including certificates, mint,
outputs, scripts, and witness counting.

Add Dijkstra era handling across the read library: block/tx
parsing, era value handling, tx generators, and output types.

Update CHaP index-state, flake inputs, and align all package
dependency bounds with cardano-node 10.6.2 ecosystem.