guard-tree's no-cargo guards depend on python3 + site-packages: 10 guards (16 rows) red without yaml/tomli, 21 guards (30 rows) red with no python3 — latent on every runner that differs from intel's #3697

Measured on the local 0.70 batch (release/0.70-batch-1 @ 59f7fe77a, main a877fa056 plus 9 constituents) with bash scripts/guard_tree.sh --no-cargo under four interpreter shapes, each as a python3 stub first on PATH:

Red under -S (10): check_arch_constraints_contract_covers_fallback (self-test, run), check_kind_labels_derived (s, r), check_parity_block_refusals (s, r), check_parity_receipt (r), check_pathonly_devdeps_unused_in_src (r), check_perf_receipt_fields_have_producers (s, r), check_roadmap_completion_is_cited (r), check_roadmap_diff_additive (r), check_roadmap_fragment_required (s, r), check_roadmap_ids_unique (s, r).

Additionally red with no python3 (11 more): check_cascade_clean_room_gate (r), check_complexity_ratchet (r), check_dogfood_coverage (r), check_fleet_pv_shapes_gate (s), check_milestone_cut (s, r), check_nextest_ci_profile_no_fail_fast (s, r), check_no_hosted_runners (s, r), check_ont_ratchet (s), check_reconcile (r), check_t2_preflight_verdict (s), check_thresholds_in_matrix (r).

Why this is latent, not red today: every runner that takes guard-tree today has a python3 with PyYAML. intel's 3.10 has no TOML reader, which is exactly what reddened check_pathonly_devdeps_unused_in_src on intel-clean-room-3 (run 35591815643, fixed by #3692 → UNMEASURED). A runner image that drops PyYAML, or ships no python3 (the sovereign-ci image has none by design), reds these guards for a reason no PR can fix: the per-host "flake" shape.

What lands (0.70): per guard, one of:

• (a) remove the python dependency (the root fix, the direction of parity_receipt_denominator.sh needs a python-free classify() — then drop its UNMEASURED path #3694: no python in automation). For YAML reads, prefer pv/bashrs/Rust readers the repo already builds.
• (b) where a python reader stays, classify its absence as fleet state: UNMEASURED … reason=<no-yaml|no-python|no-toml-reader>, exit 0, surfaced by guard_tree (guard_tree.sh drops a passing guard's output — an UNMEASURED exit-0 reads as PASS in every CI log, so #3567's verdict can never be observed #3651), never red. That is the ruling applied to fleet-pv (release(0.69): batch — 15 receipted PRs + fleet-pv andon fix, one CI run, one queue slot #3669's andon) and to pathonly (guard-tree is RED on every intel runner since the 0.69 batch: check_pathonly_devdeps_unused_in_src.sh exits ENV for a missing TOML reader — fleet state must be UNMEASURED #3692).

Plus one guard-tree row: guard_tree.sh --no-cargo under a python3 -S stub and a no-python3 stub. It must end 0 failed, with every python-dependent guard either python-free or printing UNMEASURED, so a new python-dependent guard can't land red-on-another-runner.

My own check_nextest_ci_profile_no_fail_fast.sh (#3626) is on the no-python3 list. It reads TOML through a purpose-built fallback on 3.10, but with no interpreter at all it reports ENV rc 2, which guard_tree counts as FAIL: same class, (b) applies.

Refs #3692 #3694 #3651 #3647 #3626