Bump Grafana Helm chart to v9.3.2

[jekader]

The Helm chart used to deploy Grafana is outdated and missing useful features and bugfixes. One of the features usefult for DevLake is the ability to mount configmaps - useful when overriding dashboard configs.

Updating it to latest stable version 9.3.2:
https://github.com/grafana/helm-charts/releases/tag/grafana-9.3.2

Also, the code of the dependency is removed - will be installed during the helm dependency update step and matched against the checksum in Chart.lock.

This fixes #327

[jekader]

CI is failing due to a timeout when waiting for the Grafana PVC to get claimed:

client.go:144: 2025-08-11 03:03:00 [debug] creating 19 resource(s)
wait.go:50: 2025-08-11 03:03:00 [debug] beginning wait for 19 resources with timeout of 40m0s
ready.go:284: 2025-08-11 03:03:00 [debug] PersistentVolumeClaim is not bound: default/deploy-test-grafana
ready.go:284: 2025-08-11 03:03:02 [debug] PersistentVolumeClaim is not bound: default/deploy-test-grafana
ready.go:284: 2025-08-11 03:03:04 [debug] PersistentVolumeClaim is not bound: default/deploy-test-grafana

the kind cluster does have a storage class:

NAME                 PROVISIONER             RECLAIMPOLICY   VOLUMEBINDINGMODE      ALLOWVOLUMEEXPANSION   AGE
standard (default)   rancher.io/local-path   Delete          WaitForFirstConsumer   false                  28s

Why does it not work? Maybe, the PVC request is larger than the disk on the node? The defaults have not changed across releases so maybe this is some internal bug within the CI setup.

[jekader]

CI should be passing now, issues fixed and tested in my fork

[klesh]

@matrixji @JorgeGar Can you help reviewing this PR? Thanks.

[klesh]

Thanks for your contribution.
Let's move things forward