Hybrid/headless crawl path is too slow

Hybrid mode spends most of its time on the browser critical path for work that does not help discovery.

Main costs today:

• Fetch intercepts every resource (*), then body-fetch + parse before continue. Images/fonts/media stall the page.
• Only one browser is used. -c is ignored because concurrent tabs on one Chrome break under CDP (see #1215 / #1488).
• Default page-load strategy is heuristic (WaitStable), which hangs on SPAs with continuous network activity.