Update the runner extension to v0.4.6. (#22)

Improves the performance of the filters by ~25%.
Reduces memory consumption of the filters & removes the need to
increase the CODAL stack size.

pxt.json

@@ -4,7 +4,7 @@
     "description": "Machine learning extension to support micro:bit CreateAI",
     "dependencies": {
         "core": "*",
-        "machine-learning-runner": "github:microbit-foundation/pxt-microbit-ml-runner#v0.4.4"
+        "machine-learning-runner": "github:microbit-foundation/pxt-microbit-ml-runner#v0.4.6"
     },
     "files": [
         "README.md",