source-hubspot-native: retry failures to fetch batches with associations

HubSpot APIs will sometimes return random 500 errors, or timeouts. The connector
often needs to string together thousands of successful API calls to work through
a delayed fetch of a time window, so these random errors causing a connector
restart results in a lot of downtime and wasted work.

This adds some retry logic to retry these requests a few times before crashing
the connector.

source-hubspot-native/source_hubspot_native/api.py

@@ -487,9 +487,19 @@ async def _do_batch_fetch(batch: list[tuple[datetime, str]]) -> Iterable[tuple[d
         # Enable lookup of datetimes for IDs from the result batch.
         dts = {id: dt for dt, id in batch}
 
-        documents: BatchResult[CRMObject] = await fetch_batch_with_associations(
-            log, cls, http, with_history, object_name, [id for _, id in batch]
-        )
+        attempt = 1
+        while True:
+            try:
+                documents: BatchResult[CRMObject] = await fetch_batch_with_associations(
+                    log, cls, http, with_history, object_name, [id for _, id in batch]
+                )
+                break
+            except Exception as e:
+                if attempt == 5:
+                    raise
+                log.warning("failed to fetch batch with associations (will retry)", {"error": str(e), "attempt": attempt})
+                await asyncio.sleep(attempt * 2)
+                attempt += 1
 
         return ((dts[str(doc.id)], str(doc.id), doc) for doc in documents.results)