iter-138: navigation truthfulness II — HTTP status, SPA history, honest timeouts

[ractive]

Summary

• Theme B regression fix (priority): back/forward/reload no longer hard-fail with a Timeout on same-document history traversal (SPA pushState/popstate) — iter-130 gave these commands navigate's document-commit wait, but a same-document popstate fires no fresh document-commit event, so the wait could never resolve. Same-document completion is now detected directly (readyState complete + location.href changed) via a replay-guarded probe (probe_same_document_commit_safe) instead of waiting for a commit that never arrives.
• Theme A: navigate now reports the main document's real HTTP status (404/503 pages previously returned an untruthful success envelope with no way to detect the failure short of a follow-up network call). Sourced by streaming network-event via start_daemon_stream/stop_daemon_stream alongside the existing document-event wait on the default daemon path — the daemon does not forward network-event to a plain watchResources subscriber, so this required daemon-stream plumbing, not just a subscribe.
• Theme C: same-page fragment navigation (#frag) now succeeds instead of burning the full --timeout and reporting failure (same same-document root cause as Theme B).
• Theme D: timeout error messages report real wall-clock elapsed time instead of an internal sub-budget (previously ~3× under the actual elapsed time).
• Theme E: back/forward/reload gain a real --no-wait escape hatch — previously recommended in the Theme B/C error text but rejected as an unrecognized argument.
• Theme F: back/forward/reload always re-resolve committed_url via a top-level eval location.href instead of trusting a document-event that may belong to a subframe (fixed a case where back/forward reported a cross-origin CDN subframe URL as the result).
• Theme G: navigate --with-network no longer drops committed_url/ready_state — both are now present alongside network data (and now status too).

Test plan

• cargo fmt — clean
• cargo clippy --workspace --all-targets -- -D warnings — clean
• cargo test --workspace -q — 820+ passed, 0 failed
• FF_RDP_LIVE_TESTS=1 cargo test -p ff-rdp-cli --test live live_138 — all 7 live_138_* tests pass against real headless Firefox through the default daemon path (no --no-daemon):
  • live_138_navigate_reports_404
  • live_138_navigate_reports_200
  • live_138_pushstate_back_succeeds
  • live_138_fragment_navigate_succeeds
  • live_138_timeout_message_matches_wall_clock
  • live_138_back_forward_committed_url_is_top_frame
  • live_138_with_network_keeps_envelope
• cargo run -p xtask -- check-iteration-ready --plan kb/iterations/iteration-138-navigation-truthfulness-2.md --base origin/main — 10/10 sub-checks PASS

🤖 Generated with Claude Code## Claims vs code
<generated 2026-08-10T12:19:47Z by ralph-loop>

• the → ✅ matched in diff