A run whose baseline store cannot be opened no longer dies. It used to call Actor.fail(),
so the customer saw a failed run with no output at all. It now degrades: every app is published
as a first sighting, every row carries the new STATE_STORE_UNAVAILABLE warning, and the run
status message says plainly that nothing in the run is a change. This is why the Actor was
flagged under maintenance — two of two runs by people other than the author failed outright,
while every run on the author's own account succeeded.
It falls back to the run's own key-value store, the same remedy the Website Change Monitor uses,
spelled the same way so the four monitors do not each invent their own vocabulary for it.
STATE_UNAVAILABLE is now raised only when the named store and the run's own store both
refuse, which leaves no honest degraded mode.
A first run now produces rows. With onlyReportChanges on — the default — a first sighting
was treated as "nothing changed" and nothing was pushed, so a customer's first run ended with an
empty dataset. That contradicted this README ("every app is a baseline row") and the charge
tiering, which prices a baseline as a check because "the customer got a starting point out of
it". Caught by running the deployed Actor against the listing's own example input, which
reported 4 requested and 0 rows. A first sighting is now published; a genuinely quiet check
still publishes nothing.
