feat(conga,dev): worker and scheduler reports, summer dev runs serve

- queue:work and schedule:run print a startup summary and a shutdown
  report on a terminal, log them as records otherwise, and return nil
  after a clean stop
- schedule:run --once ends with 'Ran N due entries (M failed)'
- summer dev runs bin/<app> serve with the forwarded global options,
  reports through Output, stops the old server before starting the new
  one and prints a dev report on Ctrl+C (exit 0)
- lagoon.DatabaseLabel describes a DSN without its secrets
- Long help and examples for queue:work, schedule:run, dev, docs:serve

modules/conga/README.md

@@ -224,8 +224,8 @@ queues:
 
 | Command | Arguments and flags | Description |
 |---------|---------------------|-------------|
-| `queue:work` | `--queue <name>`, repeatable | Runs a job worker in the foreground on the named queues (default: every known queue) until SIGINT or SIGTERM, then stops it within 10 seconds. An unknown queue is an error that lists the known ones. |
-| `schedule:run` | `--once` (bare) | Without `--once`: runs a worker on the `scheduled` queue that carries every plugin's periodic jobs, prints `scheduler started` and runs until SIGINT or SIGTERM, then stops within 10 seconds. With `--once`: runs, in-process and without River, every entry due in the current minute of `app.timezone` (a daily entry at its hour and minute; `Every(d)` of a minute or less on every run, a longer one when the minutes since midnight are a multiple of `d`), printing `Running scheduled command: <command> <args>` per entry or `No scheduled commands are ready to run.`. An unregistered command prints a warning and is skipped. The exit status is the first command error, after every due entry ran. There is no overlap lock: two runs in one minute run the due entries twice, as Laravel does. System cron: `* * * * * cd /app && ./bin/app schedule:run --once`. |
+| `queue:work` | `--queue <name>`, repeatable | Runs a job worker in the foreground on the named queues (default: every known queue) until SIGINT or SIGTERM. It prints a startup summary on a terminal (environment, queues with their concurrency, database host, whether the `scheduled` queue runs here) and logs it as a `startup summary` record of component `queue` otherwise. On a signal it stops the worker within 10 seconds, prints and logs a `shutdown report` (jobs done, failed and handed back to the queue, uptime, stop errors) and exits 0, or 1 when the worker fails to stop. An unknown queue is an error that lists the known ones. |
+| `schedule:run` | `--once` (bare) | Without `--once`: runs a worker on the `scheduled` queue that carries every plugin's periodic jobs, prints a startup summary (entry count and the next due entry; a `startup summary` record of component `schedule` off a terminal) and runs until SIGINT or SIGTERM, then stops within 10 seconds, prints a shutdown report (scheduled commands run and failed, uptime) and exits 0. With `--once`: runs, in-process and without River, every entry due in the current minute of `app.timezone` (a daily entry at its hour and minute; `Every(d)` of a minute or less on every run, a longer one when the minutes since midnight are a multiple of `d`), printing `Running scheduled command: <command> <args>` per entry or `No scheduled commands are ready to run.`. An unregistered command prints a warning and is skipped. It ends with `Ran N due entries (M failed)`; the exit status is the first command error, after every due entry ran. There is no overlap lock: two runs in one minute run the due entries twice, as Laravel does. System cron: `* * * * * cd /app && ./bin/app schedule:run --once`. |
 | `queue:clear` | `[queue]` (default `default`) | Deletes the available, scheduled and retryable jobs of one queue in batches until none are left and prints `Cleared N jobs`. Running jobs are never touched. |
 
 ## Dependencies