Fix Functional Test Setup

[jvoravong]

Description:

• Added a work around step to the functional test setup function to use a new client to load newly deployed CRDs.
  • The Helm cli invalidates the discovery client cache more often than the Helm sdk, I made it so our use of the Helm sdk invalidates the cache appropriately like the Helm cli. Also started drafting a Helm sdk upstream fix for this.
  • This work around gets us closer to how customers use our product and we can likely apply this fix to other CRDs (cert manager, target allocator, prometheus) installed by our functional tests.
• These changes fix only the no matches for kind "Instrumentation" in version "opentelemetry.io/v1alpha1" CRD errors specifically in our functional test code and CI/CD workflows.
• Added a value section called instrumentation.jobs.startupapicheck that would be needed to configure the job image used.
  • We need to add some waits using K8s jobs and Helm hooks for a smallest fix, we can move away from this the future though in favor of a solution based on OTL-3190 Move operator CRD installation to the crds/ folder via a subchart to resolve Helm install ordering issues #1561.
  • Customers need to be able to control if this new job is run or not with instrumentation.jobs.startupapicheck.enabled
  • Customers need to be able configure what image is used.
  • In the future we will likely need to add a instrumentation.jobs.cleanupjob no matter what that is similar to this one.
• After these changes functional tests still need to be updated for new golden file values or other recent breaking changes.