Fix OR vs AND logic error in get_model

[aymuos15]

Description

What is this PR

• Bug fix
• Addition of a new feature
• Other

Why is this PR needed?

The original condition in get_model — if existing_model is not None or network_depth is None — has a logic error. When both existing_model and network_depth are None, the or evaluates to True (because network_depth is None), causing load_model(None) to be called, which crashes with an unhelpful Keras error.

What does this PR do?

• Replaces the compound or condition with a clear if / elif / else chain so that existing_model always takes precedence when provided
• Adds an explicit ValueError guard when neither existing_model nor network_depth is supplied
• Updates the docstring to clearly describe parameter precedence rules
• Adds tests for the existing-model-takes-precedence path, the depth-only build path, and the both-None error case

References

None.

How has this PR been tested?

Three new unit tests added to tests/core/test_unit/test_classify/test_tools.py:

• test_get_model_existing_takes_precedence — verifies load_model is called (not build_model) when both arguments are provided
• test_get_model_builds_with_depth_only — verifies build_model is called when only network_depth is given
• test_get_model_no_model_no_depth — verifies ValueError is raised when both are None

All 5 tests pass locally. get_model now has 100% line coverage.

Is this a breaking change?

No. The only behavioral change is for the (None, None) case, which was already broken (crashed with an unhelpful Keras error). It now raises a clear ValueError instead.

Does this PR require an update to the documentation?

No. This is an internal bug fix with no user-facing API changes.

Checklist:

• The code has been tested locally
• Tests have been added to cover all new functionality (unit & integration)
• The documentation has been updated to reflect any changes
• The code has been formatted with pre-commit