Upgrade to latest helm dependency

[lblackstone]

Proposed changes

Related issues (optional)

[github-actions]

Does the PR have any schema changes?

Found 2 breaking changes:
Resource "kubernetes:helm.sh/v2:Chart" missing
Type "kubernetes:helm.sh/v2:FetchOpts" missing
No new resources/functions.

[rquitales]

Just added a commit to bump k8s to v1.27 for the test cluster. It seems like the client libraries being upgraded here is causing some diffing issues when the test cluster used is 1.24.13-gke.2500 resulting in the failing test. I haven't done any additional tests to see if it passes on v1.25, or v1.26 either, but GKE v1.27 works. I have also not tested if this is an issue with OSS Kubernetes, or GKE.

[rquitales]

Did more testing, and the test fails on OSS Kubernetes (KinD) and not just GKE.

Tested:

• v1.23.17 (EOL 2023-02-28): FAIL
• v1.24.15 (Latest v1.24.x): FAIL
• v1.24.13: FAIL
• v1.24.1: FAIL

---

• v1.25.11: PASS
• v1.26.6: PASS
• v1.27.3: PASS

Looks like this upgrade is causing a regression for all clusters with v1.24 or lower. @lblackstone What do you think about landing this change?

This PR causes a regression in older clusters.

[lblackstone]

What do you think about landing this change?

Lets hold off for now. This change isn't required at this time, so let's figure out what's causing the regression and we can ideally wait until it's fixed upstream.

[lblackstone]

I believe that I identified and fixed the problem causing the test failures. The updated client-go dependency no longer includes dry-run verifier support, and my initial change erroneously changed the support check to use the wrong verifier flag. As a result, the supportsDryRun function always returned false, so the Diff used inputs rather than the Server-side dry run. There is a known bug with input diffing that fails to detect drift, so the test was failing for this reason.

I updated the dry-run support check to require cluster version of v1.13+ rather than querying the cluster, and verified that the test passes again.

This was reviewed prior to latest changes that addresses the dry-run logic change.

[lblackstone]

We decided to move this change to the v4 branch, which already drops support for clusters older than v1.13. I'll rework the PR for that.

[lblackstone]

@rquitales Mind giving this another look?