feat(signals): operational scouts, starting with the follow-up scout on by default

Problem Every canonical scout today watches the customer's product: errors, replays, flags, experiments, revenue. signals-scout-inbox-validation is different in kind. It watches the self-driving system itself: it re-measures resolved reports after a soak window and files a report when a fix did not hold. It is also the natural lane for running agent checks attached to reports (see #99783, scheduled follow-up checks on a report). It is treated like any specialist scout, which is the wrong posture for something a project needs in order to be successful with self-driving: • It may not be on. register_missing_configs (products/signals/backend/scout_harness/config_registry.py) seeds a canonical scout enabled only when it is in the enabled_skills allowlist of the signals-scout flag payload, or when no allowlist is set. Whether it is enabled on a team depends on the payload at the time that team was seeded. • The posture is forward-only. Existing config rows are never re-…