Fix circular import error when Kubernetes executors are dynamically loaded with Sentry

[kaxil]

Resolves circular import that occurs when KubernetesExecutor is dynamically loaded during Airflow startup. The issue manifests as 'cannot import name from partially initialized module' errors during scheduler/triggerer initialization.

Root cause: The import chain executor → version_compat → BaseOperator → taskinstance → sentry → executor creates a circular dependency when sentry is enabled, as sentry initialization attempts to load the executor while it's already being imported.

Changes:

• Remove BaseOperator and BaseSensorOperator from version_compat.py module-level imports
• Add conditional imports directly in operator/sensor files that need these classes
• Maintain compatibility across Airflow 2.x and 3.x versions

This prevents the circular import by breaking the chain at version_compat, ensuring sentry initialization doesn't trigger recursive executor loading.

Fixes #55810

---

To Reproduce:

❯ docker run --rm apache/airflow:2.11.0 bash -c 'pip install sentry-sdk apache-airflow-providers-cncf-kubernetes==10.6.1
export AIRFLOW__CORE__EXECUTOR="KubernetesExecutor"
export AIRFLOW__SENTRY__SENTRY_ON="True"
export AIRFLOW__SENTRY__SENTRY_DSN="https://fake@fake.ingest.sentry.io/fake"
export AIRFLOW__DATABASE__SQL_ALCHEMY_CONN="postgresql://fake:fake@fake:5432/Sentry enabled (sentry_on=True)"

python3 -c "import airflow.providers.cncf.kubernetes.executors.kubernetes_executor"
'

Error log:

Collecting sentry-sdk
  ...

/home/airflow/.local/lib/python3.12/site-packages/airflow/metrics/base_stats_logger.py:22 RemovedInAirflow3Warning: Timer and timing metrics publish in seconds were deprecated. It is enabled by default from Airflow 3 onwards. Enable timer_unit_consistency to publish all the timer and timing metrics in milliseconds.
Traceback (most recent call last):
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/utils/module_loading.py", line 42, in import_string
    return getattr(module, class_name)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: partially initialized module 'airflow.providers.cncf.kubernetes.executors.kubernetes_executor' has no attribute 'KubernetesExecutor' (most likely due to a circular import)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/providers/cncf/kubernetes/executors/kubernetes_executor.py", line 43, in <module>
    from airflow.providers.cncf.kubernetes.pod_generator import PodGenerator
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/providers/cncf/kubernetes/pod_generator.py", line 49, in <module>
    from airflow.providers.cncf.kubernetes.version_compat import AIRFLOW_V_3_0_PLUS
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/providers/cncf/kubernetes/version_compat.py", line 41, in <module>
    from airflow.models import BaseOperator  # type: ignore[no-redef]
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/models/__init__.py", line 79, in __getattr__
    val = import_string(f"{path}.{name}")
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/utils/module_loading.py", line 39, in import_string
    module = import_module(module_path)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/importlib/__init__.py", line 90, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/models/baseoperator.py", line 83, in <module>
    from airflow.models.mappedoperator import OperatorPartial, validate_mapping_kwargs
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/models/mappedoperator.py", line 54, in <module>
    from airflow.triggers.base import StartTriggerArgs
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/triggers/base.py", line 27, in <module>
    from airflow.models.taskinstance import SimpleTaskInstance
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/models/taskinstance.py", line 106, in <module>
    from airflow.sentry import Sentry
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/sentry.py", line 196, in <module>
    Sentry = ConfiguredSentry()
             ^^^^^^^^^^^^^^^^^^
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/sentry.py", line 88, in __init__
    executor_class, _ = ExecutorLoader.import_default_executor_cls(validate=False)
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/executors/executor_loader.py", line 314, in import_default_executor_cls
    executor, source = cls.import_executor_cls(executor_name, validate=validate)
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/executors/executor_loader.py", line 302, in import_executor_cls
    return _import_and_validate(executor_name.module_path), executor_name.connector_source
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/executors/executor_loader.py", line 286, in _import_and_validate
    executor = import_string(path)
               ^^^^^^^^^^^^^^^^^^^
  File "/home/airflow/.local/lib/python3.12/site-packages/airflow/utils/module_loading.py", line 44, in import_string
    raise ImportError(f'Module "{module_path}" does not define a "{class_name}" attribute/class')
ImportError: Module "airflow.providers.cncf.kubernetes.executors.kubernetes_executor" does not define a "KubernetesExecutor" attribute/class