Test the rotation callbacks against the real rotation plan

[DominicOram]

A number of the issues found in #463 were not caught because the tests in test_rotation_callbacks are run against a fake_rotation_scan rather than the real one, which means that we have no tests covering the communication between the real rotation plan and the real callbacks. #477 adds a couple but it would be better if test_rotation_callbacks used the real plan.

Acceptance Criteria

• test_rotation_callbacks tests against the real plan (using the simulator should still be a good test)
• New tests added in Hotfixes found from testing on the beamline #477 are removed if they are now covered