Raise when thermocycler hold time is unavailable

[Abdeltoto]

Summary

• Raise RuntimeError in the Opentrons HTTP thermocycler backend when hold time is unavailable because no profile is running.
• Let is_profile_running() propagate backend RuntimeErrors instead of converting them to False.
• Add tests for generic profile-running error propagation and the Opentrons HTTP backend behavior.

Fixes #635

Test plan

• python -m ruff check pylabrobot/thermocycling/thermocycler.py pylabrobot/thermocycling/opentrons_backend.py pylabrobot/thermocycling/thermocycler_tests.py pylabrobot/thermocycling/opentrons_backend_tests.py
• python -m pytest pylabrobot/thermocycling/thermocycler_tests.py::ThermocyclerTests::test_is_profile_running_raises_if_no_profile_running
• python -m pytest pylabrobot/thermocycling/thermocycler_tests.py::ThermocyclerTests::test_is_profile_running_preserves_not_implemented_error
• python -m pytest pylabrobot/thermocycling/opentrons_backend_tests.py::TestOpentronsThermocyclerBackend::test_get_hold_time_raises_if_not_running skipped locally because ot_api is not installed.

[rickwierenga]

thanks for the PR! one small comment/ suggestion

[rickwierenga]

do you want to raise on RuntimeError here as well? False seems incorrect

[Abdeltoto]

Thanks, that makes sense. I updated this so \is_profile_running()\ now propagates the \RuntimeError\ instead of returning \False\, and adjusted the test accordingly.

[rickwierenga]

Let's maybe keep the return values in the chatterbox until we have a proper simulator for it. This breaks CI and potentially existing protocols.

[Abdeltoto]

Good point. I reverted the Chatterbox behavior and kept its existing 0.0 return for now, so this PR doesn’t affect simulator-style usage or existing protocols. The runtime error behavior is now limited to the Opentrons backend, and I updated the tests/PR summary accordingly.

[rickwierenga]

thanks!