fix: fail fast on deploy failure in Anaconda jobs

The Anaconda test job in the ironic-tempest-plugin could enter a
"deploy failed" state and continue waiting for the node to reach
the next state, leading to unnecessary delays.

This issue was caused by a read timeout from the remote mirror, a
common occurrence, but the test should respond by failing fast to
improve efficiency.

Changes made:
- Added the `abort_on_error_state` variable to the
`wait_for_agent_heartbeat` method in
`ironic_tempest_plugin/common/waiters.py` to enable early termination
when a node enters an error state.

- Updated logic in
`ironic_tempest_plugin/tests/scenario/baremetal_manager.py` to utilize
the new variable for consistency across test scenarios.

Fixes: https://bugs.launchpad.net/ironic/+bug/2075359
Change-Id: Ied60ae203a838e4a19fc10281f680dbb97908162
2 files changed