feat(crons): Add Solid Queue recurring task monitoring

[wedamija]

Sends Crons check-ins for Solid Queue recurring tasks (config/recurring.yml and dynamic tasks). Each run sends in_progress/ok/error check-ins with the task key as the slug and the task's schedule (crontab + IANA time zone) as the monitor config, so Sentry upserts the monitor.

Enable it like the other scheduler integrations (:sidekiq_cron, :rufus_scheduler):

Sentry.init do |config|
  config.enabled_patches << :solid_queue_recurring
end

How it works. RecurringTask#enqueue (public, same signature in Solid Queue 1.1 through 1.7) marks the job it enqueues, and the job's serialized payload carries the slug and schedule to the worker, where an around_perform sends the check-ins. The schedule is resolved in the scheduler process, using the same zone Solid Queue uses: the zone in the schedule, then SolidQueue.time_zone (1.7+), then the local zone. The slug comes from the task key, not the job class, so one job class behind several tasks gets one monitor per task. Retries keep the payload, so a retried run checks in again. This also works when a task's job uses a non-Solid Queue adapter.

What is skipped.

• class: and command: tasks are both monitored.
• Schedules Sentry would read differently (seconds, L, #, day-of-month AND weekday, non-IANA zones) still send check-ins, but without a monitor config. The conversion is the shared Sentry::Cron::FugitSchedule from ref(crons): Extract Fugit schedule conversion #3139.
• Jobs that already use Sentry::Cron::MonitorCheckIns (sentry_monitor_check_ins) are left alone.

No CHANGELOG entry: the changelog is generated from PR titles (changelog.policy: auto) and has no Unreleased section.

Stacked on #3139 (which is stacked on #3110).