[Pipelines] Add Unknown to retryable statuses (mlrun#7231)

Currently we only retry on a stable pipeline status (a terminal state
basically), but an "Unknown" state can possibly also be terminated, and
there is no reason to not allow retrying on it.

This PR creates the `retryable_statuses` list, and a followup will use
this list in
[here](https://github.com/mlrun/mlrun/blob/916f8e006b04216883bae9f8487d223935eb7554/server/py/services/api/crud/pipelines.py#L287).

pipeline-adapters/mlrun-pipelines-kfp-common/setup.py

@@ -21,7 +21,7 @@
 
 setup(
     name="mlrun-pipelines-kfp-common",
-    version="0.3.7",
+    version="0.3.8",
     description="MLRun Pipelines package for providing KFP common functionality",
     author="Yaron Haviv",
     author_email="yaronh@iguazio.com",

pipeline-adapters/mlrun-pipelines-kfp-common/src/mlrun_pipelines/common/models.py

@@ -111,3 +111,9 @@ def transient_statuses():
             for status in RunStatuses.all()
             if status not in RunStatuses.stable_statuses()
         ]
+
+    def retryable_statuses(self):
+        return self.stable_statuses() + [
+            RunStatuses.unknown,
+            RunStatuses.runtime_state_unspecified,
+        ]