Owner notices: the stall, one webhook path from env to Discord (#171)

Rollout note: the owner_notices table starts empty, so the first pass after deploy sends for conditions already true — correct per one-row-per-episode; say so rather than have it reported as a bug. Prod step: create the webhook, set DISCORD_WEBHOOK_URL on the deployment, redeploy — unset is silent by design, and without that step the feature ships dark. Security invariants preserved: the webhook address is a secret in the class of TOKEN_KEY (never logged, never on a config-printing line), and the owner gate is unchanged.

backend/internal/store/migrations/0020_owner_notices.sql

@@ -0,0 +1,12 @@
+-- Owner notices (issue #171): one row per episode, remembered only as "the
+-- owner was told". The row's presence is the whole state — the poller checks
+-- it before sending, writes it after a successful send, and clears it when
+-- the condition no longer holds. site is '' for a fault that is not one
+-- Site's; the composite primary key is what makes the row a lock against a
+-- second message for the same episode.
+CREATE TABLE owner_notices (
+	condition text NOT NULL,
+	site text NOT NULL,
+	notified_at bigint NOT NULL,
+	PRIMARY KEY (condition, site)
+);