fix: bump to madwizard 0.17.0 to pick up load time optimizations

This gives us support for pre-generating the parsed ASTs for the guidebooks, and loading those instead of doing a full parse from source every time a user runs a guidebook.

This should happen transparently, by virtue of running `npm run mirror`

Author: starpit

package-lock.json

@@ -23,6 +23,6 @@
     "access": "public"
   },
   "dependencies": {
-    "madwizard": "^0.16.0"
+    "madwizard": "^0.17.1"
   }
 }