feat(sync): apply the lab rejection reason map on a schedule

[amitdugar]

The labs' half of this is already automatic: lab-metadata-sender.php runs
inside sync-sts every five minutes, so a lab that upgrades starts sending its
reason tables within minutes and the receiver records what each of its ids
means. Only the last step -- putting those answers onto samples already stored
-- needed running by hand, and nobody should have to remember it.

Scheduled on STS instances only. The map exists nowhere else, and a lab's own
ids are correct on the lab: rewriting them there would break its own reports.

Not a run-once. RunOnceUtility logs a script as executed on any successful
completion, so one firing at upgrade would find an empty map -- no lab has
synced yet at that moment -- log itself done, and never run again while the
answers arrive over the following weeks.

Two things the nightly run needed:

• --quiet, so it speaks only when it actually rewrote something. A job that
  reports "nothing to do" every night teaches everyone to ignore its output.
• an explicit check that the map table exists. Without it, an STS mid-upgrade
  threw and the global handler exited 0: a scheduled job failing invisibly,
  which is the failure mode this whole line of work has been about.

Verified against a copy of a country's database: a seeded mapping reported
"vl lab 86: reason 27 -> 12 (644 rows)" and --dry-run left all 644 untouched.