Genie.py ruff & pyright

[jackbdoughty]

• makes genie.py pass ruff & pyright (when running only pyright genie.py)
• changes a few of other files though so some files that this PR makes changes to which are already passing ruff & pyright may need revisiting

[jackbdoughty]

Failing checks are from files that are to be covered in other PRs afaik

To resolve the cyclic import issue, we can break the dependency between genie_python.genie_epics_api and genie_python.genie. Since the imports under TYPE_CHECKING are only used for type hints, we can use forward references (string literals) for type annotations instead of importing the actual symbols. This eliminates the need for the TYPE_CHECKING block and removes the cyclic dependency.

Specifically:

1. Replace the TYPE_CHECKING block (lines 33-39) with forward references in the type annotations where PVValue, _CgetReturn, _GetbeamlineparsReturn, and _GetSampleParsReturn are used.
2. Remove the import of TYPE_CHECKING from the typing module, as it will no longer be needed.

---

To resolve the cyclic import issue, we can break the dependency loop by moving the type definitions (_CgetReturn, PVValue, etc.) into a separate module. This new module will serve as a shared location for type definitions, which can then be imported by both genie_python.genie and genie_python.genie_simulate_impl without creating a circular dependency. This approach adheres to the recommendation of moving interdependent parts into a third module.

Steps to fix:

1. Create a new module, e.g., genie_python.genie_types, to house the type definitions.
2. Move the type definitions (_CgetReturn, PVValue, etc.) from genie_python.genie to genie_python.genie_types.
3. Update the TYPE_CHECKING block in genie_python.genie_simulate_impl to import these types from genie_python.genie_types instead of genie_python.genie.
4. Update any other references to these types in the codebase to use the new module.

---

To resolve the cyclic import issue, we can break the dependency by moving the PVValue type annotation to a string literal. This approach avoids importing PVValue entirely, as type annotations can be expressed as strings in Python. This eliminates the cyclic dependency while preserving the type information for static analysis tools.

---

[Tom-Willemsen]

Happy for this to be merged once pyright gives a green tick (which may just be merges/rebase against branches mentioned where these errors were fixed?)

To fix the problem, we should remove _CgetReturn from the import statement on line 34. This will eliminate the unused import while preserving the other imported items that might still be in use. This change will not affect the functionality of the code, as _CgetReturn is not referenced anywhere in the provided snippet.

---