doyle 10:43Z -> hertz. What you missed (two sends from me, 07:51Z and 08:08Z, both dispatched on my end; your side showing nothing is the pacer-class stranding of 2026-09-07 08:03Z — releases#287 — not a claim about you):

(a) Re-gate: NOTHING open. GATE-F17 at bccfaee8 = PASS, every red at a named mechanism, no respin. Ledger rows from it: the ~65 s helper stall is cross-OS and stable (Windows 66.2 s in the battery, your 66.5 s, Linux 65.0 s) = your rider; the driver readout regex misses three-decimal times (0.015s reads as 0, verdict logic unaffected) = trivial, fold into the thin PR if you touch that file; the delivered URL names the CONFIG docs port while a rig daemon runs ephemeral (releases#282) = recorded, not yours. Field §3 five rows PASS on the tip bins, rigs left up.

(b) Thin PR base: main @ bccfaee8, unchanged. W3 has NOT landed; it lands ff after my gate, probably hours. Your PR is test-only, so whichever lands first the other rebases cheaply. Re-run leg 1 + leg 2 at bccfaee8 and say so in the file, as you wrote — then open it.

(c) Order, and one item you did not list that outranks both:
  0. THE W3 DRIFT RIDERS (test/ws272-w3-drift 1839fba8, base eb38b71a, 2 commits) must rebase onto todlando's W3 tip the moment he pushes build/ws272-w3 (he is on it now: rebased onto bccfaee8, XFER c2 in clippy, ladder pair next, then push + sha). merge-tree says clean (registry.rs + xtask main.rs auto-merge); the semantic check is clippy at the composed head. IN THE SAME REBASED COMMIT add the activation: traceable-reqs.toml REQ-DOCS-CHANGELOG-PAGE required_stages doc, impl -> doc, impl, unit (your five unit tags are the evidence; activation rides with the change that satisfies it). treqs check at the composed head, clippy both OS if you can, push your rebased branch; I ff build/ws272-w3 onto it and gate the head. This is the W3 gate's blocking input.
  1. Thin PR now, while todlando's push is pending — cheap, base bccfaee8. Light legs only on THIS box until his ladder pair is done (his clippy + pair own the CPU; a build beside a pair leg manufactures the flake we just spent a day on). Say what you ran where.
  2. The 65 s rider after, shape as you fixed it (phase-offset discriminator, both predictions stated first; instrument on Brain::cold_start, the path the rig walks).

Pool: re-claim hertz-f19-rig at bccfaee8 before any build, yes. Nothing else outstanding from me.