ClassScanner - Fix guard for preboot cache scenario

[totten]

Overview

Wanted to inspect how class-scanner caches were behaving during test-runs. So I applied this hack (HitOrMiss reporting - 02106e8) and then ran some tests, eg

CIVICRM_UF=UnitTests phpunit8 tests/phpunit/CRM/Case/WorkflowMessage/CaseActivityTest.php

It showed an extremely high number of non-caching calls. You can see in the HitOrMiss report below. The patch should improve caching.

Before

All attempts to instantiate the ClassScanner::cache('index') cause it to create dummy/one-off placeholders, as per

if (empty($_DB_DATAOBJECT['CONFIG'])) {
  return new \CRM_Utils_Cache_ArrayCache([]);
}

For CaseActivityTest, it does this 13 times.

OK (3 tests, 38 assertions)
HitOrMiss: {
    "ClassScanner::cache(index) => Early usage": 13,
    "ClassScanner::cache(structure) => New arraycache": 1
}

After

Attempts to instantiate the ClassScanner::cache('index') can create a real cache, as per:

static::$caches[$name] = \CRM_Utils_Cache::create([
  'name' => 'classes',
  'type' => ['*memory*', 'SqlGroup', 'ArrayCache'],
  'fastArray' => TRUE,
]);

For CaseActivityTest, it does this -- and it only runs 1 time.

OK (3 tests, 38 assertions)
HitOrMiss: {
    "ClassScanner::cache(index) => create(...)": 1,
    "ClassScanner::cache(structure) => New arraycache": 2
}

[civibot]

(Standard links)

• If this is your first pull-request for CiviCRM, please browse CONTRIBUTING.md for information about the development and testing processes.
• If you are reviewing this pull-request, you may wish to consult the test sites and the Review Standards (long template, short template).

[demeritcowboy]

This seems clearly correct even though it wasn't what we were looking for.