🧪 Use flag-based classifiers @ Codecov config

This is supposed to help us separate the code coverag numbers by
context. Like MyPy vs. regular test runs.

Author: webknjaz

.codecov.yml

@@ -17,14 +17,24 @@ coverage:
       default:
         target: 100%
       lib:
+        flags:
+        - unit
         paths:
         - multidict/**/*.py
         target: 94.05%
       tests:
+        flags:
+        - unit
         paths:
         - tests/
         target: 100%
+      typing:
+        flags:
+        - MyPy
+        target: 100%
       typing-stubs:
+        flags:
+        - MyPy
         paths:
         - >-
           **/*.pyi