feat(labeler): support comma-separated JSONPath queries in DefaultQuery

[browdues]

Allows comma-separated JSONPath expressions in LabelConfig.DefaultQuery.

Each expression is evaluated independently and results are joined with -.

This enables K8s-style labels like namespace-name from a single query string: $.metadata.namespace,$.metadata.name. Cluster-scoped resources (no namespace) naturally produce just the name since the missing field returns no result and is skipped

Previously, DefaultQuery only accepted a single JSONPath. Plugins needing multi-field labels had no option — K8s resources collided when a Service and its auto-created Endpoints shared the same name.

• 2 new tests (comma-separated with namespace+name, single-field fallback)
• All 15 labeler tests pass

[JeroenSoeters]

I'd push back on this:

The field is documented as a JSONPath query expression, after this PR, it no longer is one. Comma-separated queries are a custom invention layered on top, this means:

• the semantics become somewhat surprising, everyone reading $.metadata.namespace,$.metadata.name would reasonably try to validate it as a JSONPath expression and be confused when it fails.
• JSONPath queries themselves can contain commas, strings.Split(query, ",") would break those

Also JSONPath already has a union operator. The standard syntax $['metadata']['namespace','name'] selects multiple fields. I believe we might be doing this already in AWS. The only restriction is that this operator can only select keys from the same top-level parent, it cannot select keys from different part of the tree. If we need this I suggest we switch to a slice on the interface DefaultQueries []string. It looks though that we might get away with the union operator here?