Update flow run policy retry settings to be nullable

This allows us to tell if these values were provided _explicitly_ on the run or if they can be updated with values discovered at runtime.

src/prefect/orion/orchestration/core_policy.py

@@ -269,11 +269,11 @@ async def before_transition(
 
         run_settings = context.run_settings
         run_count = context.run.run_count
-        if run_count > run_settings.max_retries:
+        if run_settings.max_retries is None or run_count > run_settings.max_retries:
             return  # Retry count exceeded, allow transition to failed
 
         scheduled_start_time = pendulum.now("UTC").add(
-            seconds=run_settings.retry_delay_seconds
+            seconds=run_settings.retry_delay_seconds or 0
         )
 
         failed_task_runs = await task_runs.read_task_runs(

src/prefect/orion/schemas/core.py

@@ -90,8 +90,8 @@ class FlowRunPolicy(PrefectBaseModel):
 
     # TODO: Determine how to separate between infrastructure and within-process level
     #       retries
-    max_retries: int = 0
-    retry_delay_seconds: float = 0
+    max_retries: Optional[int] = None
+    retry_delay_seconds: Optional[float] = None
 
 
 class FlowRun(ORMBaseModel):