Allow config actions to be applied to multiple config entities using wildcards

Problem/Motivation

This issue is spun off from #3418179: [meta] Make config actions more dynamic.

In that issue, we identified several use cases that follow a similar pattern. Such as...

Adding the same permission(s) to all user roles:

user.role.*:
  grantPermissions:
    - 'access content'

Affecting the view display of every content type in the same way -- say, by removing a component you don't want people to see:

core.entity_view_display.node.*.*:
  removeComponent: links

Changing settings on every instance of a field:

field.field.media.*.field_tags:
  setSettings:
    auto_create: false

Creating a view display (i.e., enabling a view mode) for every content type:

core.entity_view_display.node.*.foo:
  ensure_exists:
    # The contents of the view display here...

Overriding a base field for every media type:

core.base_field_override.media.*.thumbnail:
  ensure_exists:
    # The contents of the base field override here...

All of these uses have a similar need -- they have to be able to "select" the config items to change using * as a placeholder. Let's add that ability.

Proposed resolution

Right now, \Drupal\Core\Recipe\RecipeRunner::processConfiguration() very straightforwardly loops through every piece of config identified in a recipe's config:actions section.

Instead of taking each config name as given, we should run it through a transformation that first converts * to a regex (probably [a-z0-9_]+), then calls preg_grep() on the list of all config that exists in the active storage. Whatever comes out of that preg_grep() call is the actual list of config items that should have the action(s) applied.

User interface changes

None.

API changes

Yes, slightly new syntax for some use cases of recipes.