Profile and trim slow PRK.Client tests

Measure PRK.Client test execution at per-test granularity after the project reorganization. Identify repeatable slow tests and distinguish useful integration coverage from redundant matrices, benchmark-style checks, sleeps, polling, and expensive setup. Reduce routine CI cost where coverage can be preserved through focused cases, deterministic synchronization, consolidation, or opt-in categories.\n\nAcceptance criteria:\n- Capture and summarize per-suite and top per-test timings.\n- Confirm slow candidates across repeated runs or explain one-time initialization effects.\n- Prune, consolidate, or recategorize only where behavior coverage remains meaningful.\n- Re-run affected suites and report before/after timing and test counts.\n\nPull request: #2656 https://git.project-rk.com/prk/engines/pulls/2656