Refactor functionality for the SQLite driver to stop migration testing usage of the 'extra_prefix' connection option

Postponed on #3106531: Notify in Status Report that per-table database prefixes are no longer supported, and will throw errors in Drupal 10.0.

@Neslee Canil Pinto your patch is removing entire test or methods, which is not what we are trying to achieve here. What we are trying to achieve is to get rid of

      // Add the original simpletest prefix so SQLite can attach its database.
      // @see \Drupal\Core\Database\Driver\sqlite\Connection::init()
      $connection_info[$target]['prefix'][$value['prefix']['default']] = $value['prefix']['default'];

and replace with something else to be still determined.

However, it looks like at least one of the two cases is actually not relevant for the test, at all, in Drupal\Tests\migrate\Kernel\Plugin\MigrationPluginListTest.

So maybe that one yes, can be just removed.

Let's see this, as a first step to separate attaching the additional database from the constructor into a method of its own.

Getting a bit more shape, with deprecation of ::setPrefix. It's replaced by a ::addPrefix method that only works with scalar values - no longer need to pass an entire array, that would be an exception to be managed via ::attachSchema.

Stopping here now, until parent is committed.

Changed to MR workflow; working on this.

The MR contains some code taken from #3200743: Decouple identifier management from database Connection, introduce an IdentifierHandler class.

Changing issue component to 'database system' since the patch is adding API to the database layer.

Thanks for reviews, @daffie, made changes and commented on the MR.

Rerolled

Committed e67aa82 and pushed to 9.3.x. Thanks!

I'm not that wild about the no-op \Drupal\Core\Database\Connection::attachDatabase() method but it's hardly even going to be used and is marked @internal and it does allow us to make headway in removing per table prefixing without having to support extra_prefix.

Thank you! It will help to related

This caused a significant regression: #3260111: Regression AND bug: SQLite source and id map tables aren't joinable anymore, but SqlBase::mapJoinable() is unaware of this.