Conduct a performance test similar to the baseline, but with an added fault injection: a loop will be run in parallel that deletes one `pipelines-controller` pod every minute. This test aims to validate the hypothesis that `pipelines-controller` restarts can cause double-counting in the `tekton_pipelines_controller_pipelinerun_total` metric, leading to discrepancies with Tekton Results. The setup will use a single namespace and 100 concurrency for pipeline generation, mirroring the baseline. Analyze the resulting PLR and TR counts, looking for evidence of increased discrepancies compared to the baseline test.