Rename _to_sharding_plan -> to_sharding_plan (pytorch#2643)

Summary:
Pull Request resolved: pytorch#2643

Renaming the function without underscore, with expectation that other classes may import and use this method too

Reviewed By: ge0405

Differential Revision: D67229782

fbshipit-source-id: edb69140c1e9a945d71c908faab7fb7b38f41e88

torchrec/distributed/planner/planners.py

@@ -67,7 +67,7 @@
 from torchrec.distributed.utils import none_throws
 
 
-def _to_sharding_plan(
+def to_sharding_plan(
     sharding_options: List[ShardingOption],
     topology: Topology,
 ) -> ShardingPlan:
@@ -388,7 +388,7 @@ def plan(
                 best_plan = callback(best_plan)
 
             self._best_plan = best_plan
-            sharding_plan = _to_sharding_plan(best_plan, self._topology)
+            sharding_plan = to_sharding_plan(best_plan, self._topology)
 
             end_time = perf_counter()
             for stats in self._stats:
@@ -737,7 +737,7 @@ def plan(
                     best_plan = callback(best_plan)
 
                 self._best_plan = best_plan
-                sharding_plan = _to_sharding_plan(
+                sharding_plan = to_sharding_plan(
                     best_plan, self._topology_groups[group]
                 )
                 best_plans.append(sharding_plan)