Fix get_flow() with reinstantiate=True for flows without extensions

[JaySingh79]

Metadata

• Reference Issue: Fixes [BUG] Error: No extension registered which can handle flow when using 'get_flow()' with reinstantiate=True #1626
• New Tests Added: Yes
• Documentation Updated: Yes
• Change Log Entry: Fix get_flow() to handle flows without registered extensions gracefully

Details

Problem:
Calling openml.flows.get_flow(1, reinstantiate=True) (or any flow without a registered extension, e.g., openml.evaluation.EuclideanDistance or weka.SMO) raised ValueError: No extension registered which can handle flow: 1 (...).

from openml.flows.functions import get_flow
flow = get_flow(1, reinstantiate=True)  # ValueError

Root cause:
get_flow() directly accessed flow.extension which calls get_extension_by_flow(..., raise_if_no_extension=True).

Solution:

• Modify get_flow() in openml/flows/functions.py to check for extension availability via flow._extension and get_extension_by_flow(..., raise_if_no_extension=False)
• If an extension exists, reinstantiate as before (including strict_version handling)
• If no extension, log a WARNING via logging.getLogger(__name__) and return flow with model=None instead of raising
• Update docstring to document new behavior and add examples

Testing:

• Updated test_get_flow_reinstantiate_model_no_extension to assert warning and model is None instead of ValueError
• Added test_get_flow_reinstantiate_openml_native_flow for issue [BUG] Error: No extension registered which can handle flow when using 'get_flow()' with reinstantiate=True #1626 verification with flow_id=1
• Manually verified via mocked openml._backend.flow.get for openml.evaluation.EuclideanDistance, weka.SMO, and sklearn DecisionTreeClassifier (reinstantiates correctly)
• Existing non-server tests pass (test_are_flows_equal etc.)