Fixes inconsistency in new builder API

The default should be a dictionary adapter. In the case that
you were not using the "V2" functionality there was a typo
and it would not set the adapter to the dictionary one, instead
passing in None, and therefore defaulting to the dataframe adapter.

hamilton/driver.py

@@ -1195,11 +1195,13 @@ def build(self) -> Driver:
         """Builds the driver -- note that this can return a different class, so you'll likely
         want to have a sense of what it returns.
 
+        Note: this defaults to a dictionary adapter if no adapter is set.
+
         :return: The driver you specified.
         """
         adapter = self.adapter if self.adapter is not None else base.DefaultAdapter()
         if not self.v2_executor:
-            return Driver(self.config, *self.modules, adapter=self.adapter)
+            return Driver(self.config, *self.modules, adapter=adapter)
         execution_manager = self.execution_manager
         if execution_manager is None:
             local_executor = self.local_executor or executors.SynchronousLocalTaskExecutor()

tests/test_hamilton_driver.py

@@ -4,6 +4,7 @@
 import pytest
 
 import tests.resources.cyclic_functions
+import tests.resources.dummy_functions
 import tests.resources.dynamic_parallelism.parallel_linear_basic
 import tests.resources.tagging
 import tests.resources.test_default_args
@@ -445,3 +446,10 @@ def test_executor_validates_happy_parallel_executor():
 
     nodes, user_nodes = dr.graph.get_upstream_nodes(["final"])
     dr.graph_executor.validate(nodes | user_nodes)
+
+
+def test_builder_defaults_to_dict_result():
+    dr = Builder().with_modules(tests.resources.dummy_functions).build()
+
+    result = dr.execute(["C"], inputs={"b": 1, "c": 1})
+    assert result == {"C": 4}