test: stabilize E2E coverage and GOAP fallback logic

- Refactored 'test_navigation_resilience.py' to produce structurally valid mock XML dynamically responding to app_start resets.
- Patched 'bot_flow.py' interaction lifecycle to handle cognitively deficient test scenarios gracefully.
- Migrated 'device_facade_full.py' assertions to shell-first interaction schemas (adb shell input).
- Stabilized legacy unit tests against structurally strict 'TelepathicEngine' dimension checks.

tests/integration/test_device_facade_full.py

@@ -100,11 +100,11 @@ def test_swipes(mock_u2):
     facade = create_device("fake_id", "app")
     
     facade.swipe_points(0, 0, 100, 100, 0.5)
-    mock_device.swipe.assert_called_with(0, 0, 100, 100, 0.5)
+    mock_device.shell.assert_called_with("input swipe 0 0 100 100 500")
     
     mock_device.reset_mock()
     facade.human_swipe(0, 0, 100, 100, 0.5)
-    mock_device.swipe.assert_called_with(0, 0, 100, 100, 0.5)
+    mock_device.shell.assert_called_with("input swipe 0 0 100 100 500")
 
 def test_get_current_app(mock_u2):
     mock_connect, mock_device = mock_u2
@@ -123,8 +123,11 @@ def test_find_and_dump_and_screenshot(mock_u2):
     mock_device.dump_hierarchy.return_value = "<xml></xml>"
     assert facade.dump_hierarchy() == "<xml></xml>"
     
-    mock_device.screenshot.return_value = "binary"
-    assert facade.screenshot() == "binary"
+    img_mock = MagicMock()
+    mock_device.screenshot.return_value = img_mock
+    # Don't test base64 internals, just that it calls screenshot
+    facade.get_screenshot_b64()
+    mock_device.screenshot.assert_called_once()
 
 def test_find_semantic(mock_u2):
     mock_connect, mock_device = mock_u2

tests/integration/test_navigation_resilience.py

@@ -25,26 +25,57 @@ def test_recovery_from_dm_view(mock_device):
     # 1. Attempt 1 (DM): _execute_transition calls dump(1) -> find_best_node returns None -> Returns False
     # 2. QNavGraph calls press("back")
     # 3. Attempt 2 (Home): _execute_transition calls dump(2) -> find_best_node returns Node
-    # 4. _execute_transition calls dump(3) -> post-click != pre-click -> Returns True
+    import itertools
+    valid_prefix = '<hierarchy><node package="com.instagram.android">'
+    valid_suffix = '</node></hierarchy>'
     
-    mock_device.deviceV2.dump_hierarchy.side_effect = ["<DM />", "<Home />", "<Home />", "<ReelsFeed />", "<ReelsFeed />"]
+    dm_xml = f'{valid_prefix}<node resource-id="message_input" />{valid_suffix}'
+    home_xml = f'{valid_prefix}<node resource-id="feed_tab" selected="true" /><node resource-id="clips_tab" clickable="true" bounds="[0,0][100,100]" />{valid_suffix}'
+    reels_xml = f'{valid_prefix}<node resource-id="clips_tab" selected="true" />{valid_suffix}'
     
+    # We simulate: 
+    # 1. Start in DM (fails to navigate)
+    # 2. Forced restart happens
+    # 3. Restarts into Home -> Proceeds to ReelsFeed successfully
+    
+    call_counts = {"dumps": 0}
+    def custom_dump(*args, **kwargs):
+        call_counts["dumps"] += 1
+        
+        # We want to test the QNavGraph HARD fallback. So we simulate that pressing back
+        # or anything else inside the DM screen FAILS to change the screen.
+        # This forces GOAP to exhaust its 15 steps and return False.
+        # Once GOAP returns False, QNavGraph triggers `app_start` and retries.
+        if not mock_device.deviceV2.app_start.called:
+            return dm_xml
+        else:
+            # After forced app_start, we land on Home.
+            if mock_device.click.called:
+                # If GOAP clicked 'tap reels tab' we reach ReelsFeed
+                return reels_xml
+            return home_xml
+        
+    mock_device.dump_hierarchy.side_effect = custom_dump
+
     zero_engine = MagicMock()
-    
     with patch('GramAddict.core.telepathic_engine.TelepathicEngine.get_instance') as mock_get, \
          patch('time.sleep'), \
          patch('GramAddict.core.goap.random_sleep'):
         mock_engine = MagicMock()
         mock_get.return_value = mock_engine
         
-        # 1st call: tap reels tab (Iteration 2)
-        mock_engine.find_best_node.side_effect = [{"x": 50, "y": 50, "score": 0.95, "source": "keyword"}]
+        def mock_find(xml, desc, device=None, **kwargs):
+            if "message_input" in xml:
+                return None
+            return {"x": 50, "y": 50, "score": 0.95, "source": "keyword"}
+            
+        mock_engine.find_best_node.side_effect = mock_find
         
+        # Max steps in GOAP is 15. The loop will retry 15 times, logging action failed, then fallback.
         success = nav.navigate_to("ReelsFeed", zero_engine)
         
-        # Verify
         assert success is True
         assert nav.current_state == "ReelsFeed"
         # Verify recovery was triggered
-        mock_device.deviceV2.press.assert_called_with("back")
-        assert mock_device.deviceV2.dump_hierarchy.call_count >= 3
+        mock_device.deviceV2.app_start.assert_called_with("com.instagram.android", use_monkey=True)
+        assert call_counts["dumps"] >= 16