fix(deep): aggregate Consul health metric in the evidence gather

[WZ]

The final link in the chain

From the persisted traces of the #257 re-test: the orchestrator forms the right Consul hypothesis (#256) with a verifiable prediction (#257), but the evidence gather queried the bare consul_health_service_status metric — which returns one series per (node × status), mixed 0/1 — so the extractor produced zero usable observations and the keystone returned absent:

proposed a cause: impala-statestore is unhealthy …
gathered evidence  … → +0 observations          ← bare metric, no usable value
couldn't verify    impala-statestore is unhealthy …

(bd-management happened to confirm via Consul; impala kept coming up empty → wall-clock/exhausted.)

Fix

planPredictionQuery appends a Consul-aware hint to the metric-threshold gather prompt: query it aggregated for the service in the hypothesis —
max by (service_name) (consul_health_service_status{service_name="X",status="passing"}) → one clean value (1 passing / 0 failing) the keystone can verify → confirm.

Triggers only when the metric is consul_health_service_status (no false positives for ordinary metrics).

Verification

• tsc clean · 2 new tests (hint present for consul / absent for ordinary) · full suite green
• Live re-test pending — expect impala to now confirm the Consul-health cause consistently (matching bd-management)

Context

Closes the form → predict → gather → verify chain for bare-metal Consul incidents. The iteration that should make impala pass consistently. Feature still gated off.

[WZ]

Superseded by #261, which combines #258 + #259 + #260 into a single inc-7 accuracy PR.