Cardano Updates

Add diagnostic-only logging (no behaviour change) to pinpoint the
`InvalidWitnessesUTXOW` rejection seen when submitting legacy-DRep ballot
votes. Conway `voting_procedures` bodies are re-encoded by some wallets,
so a vkey witness ends up signed over a different body hash than the one
reaching the node.

Three [ballot-witness-diag] log points:
- txScriptRecovery: new diagnoseTxWitnesses() verifies every vkey witness
  against the exact body being submitted, logging any stale witness (with
  pubkey/keyhash) right before submitTx. Covers both signing paths.
- txSignUtils.mergeSignerWitnesses: logs wallet body re-canonicalisation,
  distinguishing first-signer (body adopted) from co-signer (swap skipped,
  witness may be stale).
- api/v1/signTransaction: logs when merge/rebuild changes the body hash the
  collected signatures were made over.

Co-Authored-By: Claude Opus 4.8 <[email protected]>

Signed-off-by: Jonathan Lim <[email protected]>

Signed-off-by: Jonathan Lim <[email protected]>

Signed-off-by: Jonathan Lim <[email protected]>

Signed-off-by: Jonathan Lim <[email protected]>

The mocks used jest.mock(..., { virtual: true }) on real modules. Virtual
mocks register by literal specifier, so they applied when the handler imported
e.g. @/utils/nativeScriptUtils directly but not reliably when the real
resolveWalletScriptAddress imported it — under suite pollution the real decode
ran instead of the mock, so the "canonical fallback cannot decode" test saw
"unknown error" instead of "invalid canonical cbor". Removing virtual:true makes
jest resolve by file path, applying each mock to every importer. Verified: the
full CJS suite passes across repeated --runInBand runs.

Co-Authored-By: Claude Opus 4.8 <[email protected]>

When freeUtxos.bot.test.ts runs in the full suite, three cases that
exercise the canonical-scriptCbor fallback in resolveWalletScriptAddress
fail intermittently — decodeNativeScriptFromCbor is never observed as
called and the error path produces "unknown error" instead of the
mocked message. In isolation the same tests pass.

Root cause: nativeScriptUtils.test.ts imports the real
@/utils/nativeScriptUtils (no mock), caching the module in Jest's
registry. By the time this suite declares its jest.mock(..., {virtual:
true}) replacement, the bindings inside the transitively imported
@/lib/server/walletScriptAddress are already wired to the real
implementation, so the mocks never fire.

jest.resetModules() in beforeAll forces the dynamic handler import to
walk a fresh module graph that honours this suite's mocks. Verified
with `npx jest` (full suite): 357 pass / 0 fail / 2 skipped.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

Content was already integrated via #269 (staging/preprod-prisma7-mesh2, which
merged origin/main into preprod and resolved all conflicts). That PR was
squash-merged, so main was not in preprod's ancestry and the preprod->main PR
re-conflicted. This records main as a parent without changing preprod's tree,
making the preprod->main merge clean.

cardano-node: remove iohk-monitoring tracing backend

Signed-off-by: Jonathan Lim <[email protected]>

Signed-off-by: Jonathan Lim <[email protected]>

sim-rs: fix build by migrating to spec-faithful Praos lottery API

Signed-off-by: cryptodj413 <[email protected]>

Signed-off-by: Chris Gianelloni <[email protected]>

Feed the ancillary proof input into multi-signature creation and persist the returned verifier data on the certificate.

Carry the prover input into the clerk, return the verifier data alongside the aggregate signature, and accept optional verifier data on verification.

Build the aggregate signature ancillary input from the genesis and parent certificates and store the produced verifier data on the certificate.

Wrap the STM ancillary types in ProtocolKey (bytes hex codec), add optional prover and verifier fields to Certificate, and fold them into its hash.

Add nullable ancillary columns (migration 44) and thread the data through the certificate record conversions.