Task: Duplicate event prevention/cleanup

Problem: There are apparently a variety of situations where you can end up w/ duplicate events, sometimes hundreds to thousands of them. This can pretty much perpetually bog down a site until cleaned up manually.

## 1) Prevent duplicate recurring events

I actually think this is something WP core overlooked, and should probably revisit in core trac. Single-events are prevented from being registered if there is a similar action/args event already scheduled within +/-10mins. Docs actually [say the same](https://developer.wordpress.org/reference/functions/wp_schedule_event/) about recurring events, but it's not true.

Part of this is already implemented in an upcoming release, seen here: https://github.com/Automattic/Cron-Control/pull/231/commits/bbb53aeb9d83cd753f9c584b2f38c3ce91a3992c#diff-467459221a72437e5497c04a9ac91579c53ef140b9da6bb0b67c043dc1fb0490R41-R61. If an event has the same action/args/schedule, then we're just gonna deviate from core a little and prevent registration. If multiple of the same event on the same schedule is needed for some reason, then you'd just need to add some varying args when registering them.

Perhaps could take this a step further and either re-query with SRTM (send reads to master), or have a stricter unique constraint on the table (though could be tricky w/ varying event statuses).

## 2) Cleanup existing duplicate recurring events

There are quite a lot of sites in the wild now with duplicate recurring events, some to extreme degrees with hundreds of each. On a large MS, this means cron is going to struggle unnecessarily, and just have an overall heavy load that we can prevent.

So w/ the new contract in place where we prevent duplicate recurring events, we can then go through and cleanup any existing duplicates, as well as ones that may continue to slip through the cracks due to other race conditions. I'm thinking the daily `a8c_cron_control_clean_legacy_data` internal job would be great for this.

## 3) Handle DB unique constraint errors.

Mentioned in https://github.com/Automattic/Cron-Control/issues/147, we do need to handle the cases where events run into the unique constraint. Notably, a rare but possible situation is a single event sharing the same action as a recurring event, if the timestamps collide we need to make sure to handle it gracefully.

We'll need to document this "core deviance" in the readme.