#3122 Docs: enable-managed-config should say to re-run config migrate when its push fails

What changed

swamp-club#3117 and swamp-club#2621 shipped in swamp/pull/2893 (commit f98af8ee). They change how swamp datastore config migrate behaves:

• It pulls the datastore's config/ tier before checking the migration sentinel. A repo whose cache predates another repo's migration now reports already migrated instead of migrating again.

• It writes the sentinel only after the migrated files are pushed. When a push fails, it exits non-zero with:

  The migration is saved locally but was not published to the datastore: . Run 'swamp datastore config migrate' again to publish it.

  Re-running migrate is the recovery. swamp datastore sync --push is not, because after a failed first push it would publish the files without the sentinel.

• The command now needs the datastore to be reachable even when the repo is already migrated, because of the pull.

Which manual page

content/manual/how-to/swamp-serve/enable-managed-config.md, section 2, after the idempotence example and before the concurrency warning.

Suggested content

If the push fails, migrate exits non-zero and keeps the migration locally. Run swamp datastore config migrate again once the datastore is reachable; the re-run publishes the migration. Migrate pulls the datastore's config before it checks whether the datastore is already migrated, so the datastore must be reachable even on a repo that is already migrated.

Section 3 ("Configure other instances") already says to run migrate on each instance. After the fix that is accurate even when the instance was set up before the first migration, so it needs no change.