fix: bump to madwizard 0.21.4 to pick up fix for regression in support for getting started guidebooks

we broke recently support for the old glue and any guidebooks that had a python language

```python
exec: ray-submit
--8<-- "./main.py"
```

for that kind of pattern. we weren't executing them with the python interpreter.

non-gpu4 test profile covers this. This leverages the new "Getting Started" demo guidebook from @guidebooks/store:0.7.0

Author: starpit

Diff for: .github/workflows/kind.yml

@@ -16,6 +16,7 @@ jobs:
           - non-gpu1/keep-it-simple
           - non-gpu2/keep-it-simple
           - non-gpu3/keep-it-simple
+          - non-gpu4/keep-it-simple
           - non-gpu1/ray-autoscaler
           - non-gpu1/mcad-default
           - non-gpu1/mcad-coscheduler

Diff for: package-lock.json

@@ -23,7 +23,7 @@
     "access": "public"
   },
   "dependencies": {
-    "madwizard": "^0.21.3",
-    "@guidebooks/store": "^0.6.7"
+    "madwizard": "^0.21.5",
+    "@guidebooks/store": "^0.7.1"
   }
 }

Diff for: plugins/plugin-madwizard/package.json

@@ -0,0 +1,28 @@
+{
+  "name": "keep-it-simple",
+  "creationTime": 1660657756574,
+  "lastModifiedTime": 1660658307175,
+  "lastUsedTime": 1660742079201,
+  "choices": {
+    "madwizard/apriori/use-gpu": "don't use gpus",
+    "madwizard/apriori/arch": "x64",
+    "madwizard/apriori/platform": "darwin",
+    "madwizard/apriori/mac-installer": "Homebrew",
+    "madwizard/apriori/in-terminal": "HTML",
+    "Start a new Run####Connect Dashboard to an existing Run####Boot up a Cloud Computer####Shut down a Cloud Computer": "Start a new Run",
+    "Run with CodeFlare Model Architecture####Bring Your Own Code####Demos": "Bring Your Own Code",
+    "BERT####MLFlow Demo####Tensorboard Demo": "MLFlow Demo",
+    "Location of your working directory": "{\"Location of your working directory\":\"tests/kind/inputs/qiskit-runtime-env-with-conda\"}",
+    "Provide custom base image, if any": "{\"Provide custom base image, if any\":\"rayproject/ray:1.13.1-py37\"}",
+    "AWS####IBM####My data is not stored in S3": "My data is not stored in S3",
+    "Run Locally####Run on a Kubernetes Cluster": "Run on a Kubernetes Cluster",
+    "My Cluster is Running Locally####My Cluster is Running on Kubernetes": "My Cluster is Running on Kubernetes",
+    "expand((kubectl config get-contexts -o name | grep -E . >& /dev/null && kubectl config get-contexts -o name) || (kubectl version | grep Server >& /dev/null && echo \"${KUBE_CONTEXT_FOR_TEST-In-cluster}\" || exit 1), Kubernetes contexts)": "kind-codeflare-test",
+    "expand([ -z ${KUBE_CONTEXT} ] && exit 1 || X=$([ -n \"$KUBE_NS_FOR_TEST\" ] && echo $KUBE_NS_FOR_TEST || kubectl ${KUBE_CONTEXT_ARG} get ns -o name || oc ${KUBE_CONTEXT_ARG} get projects -o name); echo \"$X\" | sed -E 's#(namespace|project.project.openshift.io)/##' | grep -Ev 'openshift|kube-', Kubernetes namespaces)####Create a namespace": "default",
+    "Number of CPUs####Number of GPUs####Minimum Workers####Maximum Workers####Worker Memory####Head Memory": "{\"Number of CPUs\":\"500m\",\"Number of GPUs\":\"0\",\"Minimum Workers\":\"1\",\"Maximum Workers\":\"1\",\"Worker Memory\":\"1.5Gi\",\"Head Memory\":\"1.5Gi\"}",
+    "Keep It Simple####Use the Ray Autoscaler####Use the Multi-user Enhanced Kubernetes Scheduler": "Keep It Simple",
+    "Run with CodeFlare Model Architecture####Bring Your Own Code####Demos": "Demos",
+    "Training Tasks####Fine Tuning Tasks": "Training Tasks",
+    "Getting Started Demo####MLFlow Demo####Tensorboard Demo": "Getting Started Demo"
+  }
+}