Cardano Updates

* test(uplc-evaluator): replace polling with fsnotify, raise timeout (#7762)

The integration-tests waited for service output via 50ms-interval
polling with a hard-coded 5s timeout. On loaded macOS Hydra builders
the polling lost time to scheduler jitter and tripped the timeout
even when the service did produce the file in time.

waitForFile now subscribes via fsnotify and uses timeout(2) only as
a wall-clock safety net (default 20s, exposed as defaultWaitMs).
The pre-watch doesFileExist check closes the race window where the
file is written before the watcher is registered.

Negative checks (file should not appear) switched from short-timeout
polling to plain doesFileExist via new resultPath/errorPath helpers,
so they no longer pay the safety-net timeout when asserting absence.

Drops the timeout parameter from waitFor*OrFail and waitForResult/
waitForError (was always called with the same value); the only
parameterised primitive left is waitForFile.

* Potential fix for pull request finding

Co-authored-by: Copilot Autofix powered by AI <[email protected]>

* test(uplc-evaluator): address Copilot review comments

- Drop unused waitForError import (was triggering -Wall unused-import).
- Rename local errorPath shadow → errorFilePath in 3 invalid-UUID test
  cases (the local binding is a String-derived path; the imported helper
  takes a UUID, so they cannot be unified — just unshadow).
- Remove stale "(5 second timeout)" comments; the timeout is now driven
  by defaultWaitMs and the comments would mislead future debugging.
- Reword waitForFile haddock: the doesFileExist probe runs *after*
  watcher registration, not before, to cover events that fired before
  we were listening.

---------

Co-authored-by: Copilot Autofix powered by AI <[email protected]>

Nix's default `.#` flake ref treats the flake as a Git-tracked input; in
bare-repo + worktree setups this produces "Path 'flake.nix' is not
tracked" errors even though the file is tracked. The `path:` prefix
sidesteps Git and reads the directory directly, with identical output
in standard clones. Encountered during Plutus 1.62.0.0 release, step 6
(`publish-gh-release`).

- Drop unused waitForError import (was triggering -Wall unused-import).
- Rename local errorPath shadow → errorFilePath in 3 invalid-UUID test
  cases (the local binding is a String-derived path; the imported helper
  takes a UUID, so they cannot be unified — just unshadow).
- Remove stale "(5 second timeout)" comments; the timeout is now driven
  by defaultWaitMs and the comments would mislead future debugging.
- Reword waitForFile haddock: the doesFileExist probe runs *after*
  watcher registration, not before, to cover events that fired before
  we were listening.

Co-authored-by: Copilot Autofix powered by AI <[email protected]>