Fix handling of min_size argument in OpenCV component (#36335)

The OpenCV image_processing component accepts a `min_size` argument in
each classifier that is then passed into the OpenCV library. The
`min_size` argument is expected to be a tuple of 2 ints, and is entered
into the platofrm schema as such. However, yaml will always produce a
list instead of a tuple, which means in practice it's impossible to use
the `min_size` argument. This changes the schema to accepts any sequence
of 2 ints that is then coerced into a tuple suitable for passing to the
OpenCV library.

homeassistant/components/opencv/image_processing.py

@@ -62,7 +62,7 @@ PLATFORM_SCHEMA = PLATFORM_SCHEMA.extend(
                             CONF_NEIGHBORS, DEFAULT_NEIGHBORS
                         ): cv.positive_int,
                         vol.Optional(CONF_MIN_SIZE, DEFAULT_MIN_SIZE): vol.Schema(
-                            (int, int)
+                            vol.All(vol.ExactSequence([int, int]), vol.Coerce(tuple))
                         ),
                     }
                 ),