chore: Rerouted legacy client row filters to data client row filters. (#1255)

**Changes made:**
- Used import aliasing to remove most of the legacy client row filter
code and use the data client row filter code in its place.
- Some of the data client row filter classes had to be slightly modified
for backwards compatibility with the legacy client row filter classes'
interface.
- As part of this move, `to_pb` has been replaced with `_to_pb`, and I
have adjusted references to `<RowFilter>.to_pb` in the libraries and
unit tests.

This is for eventually shimming up `CheckAndMutateRows`, and other row
mutation operations as well.

Author: gkevinzheng

google/cloud/bigtable/row.py

@@ -230,7 +230,7 @@ def _delete_cells(self, column_family_id, columns, time_range=None, state=None):
         else:
             delete_kwargs = {}
             if time_range is not None:
-                delete_kwargs["time_range"] = time_range.to_pb()
+                delete_kwargs["time_range"] = time_range._to_pb()
 
             to_append = []
             for column in columns:
@@ -601,7 +601,7 @@ def commit(self):
         resp = data_client.check_and_mutate_row(
             table_name=self._table.name,
             row_key=self._row_key,
-            predicate_filter=self._filter.to_pb(),
+            predicate_filter=self._filter._to_pb(),
             app_profile_id=self._table._app_profile_id,
             true_mutations=true_mutations,
             false_mutations=false_mutations,