WidgetLearnerTestMixin: Test for default learner names

Fix widgets where learner_name setting is redefined to non empty string.

Orange/widgets/model/owknn.py

@@ -27,7 +27,6 @@ class OWKNNLearner(OWBaseLearner):
     weights_options = ["Uniform", "By Distances"]
     metrics_options = ["Euclidean", "Manhattan", "Chebyshev", "Mahalanobis"]
 
-    learner_name = Setting("kNN")
     n_neighbors = Setting(5)
     metric_index = Setting(0)
     weight_index = Setting(0)

Orange/widgets/model/owrules.py

@@ -225,7 +225,6 @@ class OWRuleLearner(OWBaseLearner):
     storage_measures = ["entropy", "laplace", "wracc"]
 
     # default parameter values
-    learner_name = Setting("CN2 rule inducer")
     rule_ordering = Setting(0)
     covering_algorithm = Setting(0)
     gamma = Setting(0.7)

Orange/widgets/model/owsgd.py

@@ -59,7 +59,6 @@ class Outputs(OWBaseLearner.Outputs):
         ('Optimal', 'optimal'),
         ('Inverse scaling', 'invscaling'))
 
-    learner_name = Setting('SGD')
     #: Loss function index for classification problems
     cls_loss_function_index = Setting(0)
     #: Epsilon loss function parameter for classification problems

Orange/widgets/tests/base.py

@@ -350,8 +350,9 @@ def test_output_learner_name(self):
         new_name = "Learner Name"
         self.click_apply()
         self.assertEqual(self.widget.learner.name,
-                         self.widget.name_line_edit.text()
-                         or self.widget.name_line_edit.placeholderText())
+                         self.widget.effective_learner_name())
+        self.assertEqual(self.widget.effective_learner_name(),
+                         self.widget.name_line_edit.placeholderText())
         self.widget.name_line_edit.setText(new_name)
         self.click_apply()
         self.wait_until_stop_blocking()