Apply pytest improvements to storybook workflow (#41)

- Add parallel_workers input with cgroup-aware auto-detection
- Add smart caching for target branch results with pending marker
  pattern to avoid duplicate test runs across concurrent PRs
- Add consistent outputs for regression analysis compatibility
- Default to 6 workers for multithreaded runners, 1 for singlethreaded

Co-authored-by: Claude <noreply@anthropic.com>

Author: JamesonRGrieve