[11.x] Paginator total override

[browner12]

Resubmission to master of #46336

This allows the user to declare the total number of results a query returns. If the user provides this argument, the paginate() method will skip running its query to determine the total row count. to be clear, this does not change the number of actual results available, just how many the paginator thinks exist.

Why would we want to set this value when we can use a query to determine it?

Performance.

select count(*) as aggregate from `table_name` where `table_name`.`deleted_at` is null

can be terribly slow on large tables, depending on your database and table engine.

For example, on MySQL with an InnoDB engine, a table with 500,000 rows can take 300ms. Eliminating this query can save a considerable chunk on the entire request time.

You might ask, "why not use the simple paginator or cursor paginator to achieve this?". That would eliminate the extra query, but the downside is you lose the ability to quickly navigate directly to pages.

Users are free to calculate or choose this number any way they like. You could assume your users will never care about more than 100k records back from the most current, and use that if you like. Surprisingly, assuming your have an auto-incremented id field and rows are either never deleted or soft-deleted, you can run

select * from `table_name` order by `id` desc limit 1

to retrieve the last ID, and this query is significantly faster.