fix(helm): skip OCI repos in helm_repo_update

[NickAnge]

Summary

helm_repo_update passes all repos to helm repo update, including OCI registries (oci://). OCI registries don't use the index-based update flow — they're pulled directly by URL — so helm repo update fails with invalid reference when it encounters one.

This filters out OCI repos before calling helm repo update and returns early if no non-OCI repos remain. helm_pull already handles OCI correctly via chart_pull_path, so this only affects the update step.

Affects any chartfile using OCI repos (e.g. karpenter, cert-manager, envoy-gateway, kepler, tinkerbell in deployment_tools).

Test

• Added test_repo_update_oci_only_skips_helm — passes only OCI repos to helm_repo_update, asserts Ok(()) without needing helm on PATH
• Verified locally with rtk tool charts vendor against the deployment_tools karpenter chartfile (two OCI chart versions)

cargo test -p rtk --test charts_test test_repo_update_oci_only_skips_helm

[github-actions]

Benchmark Results

Benchmark	Test	vs tk	vs base
Diff	cluster_scoped	52.21x faster	equal
Diff	configmap_unchanged	34.03x faster	equal
Diff	deployment_nested_changes	32.88x faster	equal
Env List	Single Inline Directory (--json)	48.06x faster	equal
Env List	Single Inline File (--json)	48.04x faster	equal
Env List	All Environments (--json)	85.36x faster	equal
Eval	Single Static Environment	15.93x faster	equal
Eval	Inline Environment File	19.85x faster	equal
Eval	Eval Expression - Deeply Nested Path	16.22x faster	equal
Export (Full)	Single Static Environment	6.87x faster	equal
Export (Full)	Single Inline Environment	21.55x faster	equal
Export (Full)	All Environments	19.69x faster	equal
Export (Replace)	Single Static Environment	5.68x faster	equal
Export (Replace)	Single Inline Environment	21.46x faster	equal
Export (Replace)	All Environments	29.74x faster	equal
Helm Template	Full Export	1.42x faster	equal
Tool Importers Count	Single Lib Directory (Non-Recursive)	2.45x faster	equal
Tool Importers Count	Single Lib Directory (Recursive)	2.4x faster	equal
Tool Importers Count	Entire Lib Directory (Non-Recursive)	1.55x faster	equal
Tool Importers Count	Entire Lib Directory (Recursive)	1.18x faster	equal
Tool Importers	Global Lib File	3.47x faster	equal
Tool Importers	Single Env-Specific Lib File	1.82x faster	equal
Tool Importers	All Env-Specific Lib Files	4.79x faster	equal
Tool Importers	All Lib Files	4.91x faster	1.02x faster
Tool Importers	All Jsonnet Files	8.35x faster	equal
Tool Imports	Single Static Environment	3.49x faster	equal
Tool Imports	Single Inline Environment File	3.56x faster	equal
Validate Manifests	All Validations	-	equal
Validate Manifests	Manifest Tests Only	-	equal
Validate Manifests	Namespace Test Only	-	equal

Full results available in workflow artifacts.

Benchmark run on commit a12bb7f8e04fc7e25b4342a30dace39f77eb544b