Optimize groupb y for single partition topics

Group by operations on topics with a single partition are now optimized to avoid
creating a repartition topic. Instead, the messages are directly transformed to
use the new key, as all messages go to the same partition.

Author: ovv

quixstreams/dataframe/registry.py

@@ -71,6 +71,7 @@ def register_groupby(
         :param source_sdf: the SDF used by `sdf.group_by()`
         :param new_sdf: the SDF generated by `sdf.group_by()`.
         """
+
         if source_sdf.stream_id in self._repartition_origins:
             raise GroupByNestingLimit(
                 "Subsequent (nested) `SDF.group_by()` operations are not allowed."