[nrf fromtree] twister: Enable gathering footprint if test plan exists

If test plan is used, memory usage metrics are overridden.
Result of this operation is missing memory footprint.

This change removes redundand metrics update and fixes
described inconvenience.

Signed-off-by: Katarzyna Giądła <katarzyna.giadla@nordicsemi.no>
(cherry picked from commit bec789d364bf24d5b1a6604bc1d6cd37c06cecf3)

Author: katgiadla

scripts/pylib/twister/twisterlib/runner.py

@@ -1895,7 +1895,6 @@ def run(self):
                 except queue.Empty:
                     break
                 else:
-                    inst.metrics.update(self.instances[inst.name].metrics)
                     inst.metrics["handler_time"] = inst.execution_time
                     self.instances[inst.name] = inst
 

scripts/tests/twister/test_runner.py

@@ -2530,7 +2530,7 @@ def mock_client_from_environ(jobs):
     pipeline_q = queue.LifoQueue()
     done_q = queue.LifoQueue()
     done_instance = mock.Mock(
-        metrics={'k2': 'v2'},
+        metrics={'k': 'v2'},
         execution_time=30
     )
     done_instance.name='dummy instance'
@@ -2568,8 +2568,7 @@ def iteration_increment(value=1, decrement=False):
     assert tr.jobserver.name == expected_jobserver
 
     assert tr.instances['dummy instance'].metrics == {
-        'k': 'v',
-        'k2': 'v2',
+        'k': 'v2',
         'handler_time': 30
     }