benchmarks: avoid unreliable kubectl attach when fetching metrics

[p-se]

Fixes the benchmarks in conjunction with #5267

Additional Information

Checklist

• I have updated the documentation via a pull request in the fleet-product-docs repository.

[Copilot]

Pull request overview

This PR adjusts how benchmark metrics are fetched to avoid output duplication/collisions caused by kubectl’s attach behavior, by switching to a “create pod → wait → read logs → delete pod” flow and by enhancing the e2e kubectl wrapper to expose stdout/stderr separately.

Changes:

• Extend the e2e kubectl.Command wrapper to capture stdout and stderr independently, and add RunStdout.
• Update benchmark metrics collection to avoid kubectl run --attach and instead fetch curl output via kubectl logs after the pod completes.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
e2e/testenv/kubectl/kubectl.go	Split stdout/stderr capture; add RunStdout helper used by benchmarks.
benchmarks/record/record.go	Replace kubectl run --attach --rm metrics fetch with pod lifecycle + kubectl logs approach.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[weyfonk]

nit: in case we run into more troubles getting these logs, we could replace creating a new pod with a port forward for fetching metrics, as done in the dump CLI.

[p-se]

Hope it won't be necessary, but yeah, we may have to do that.