feat(classifier): label 'drop' via hard-rule + score threshold (-5 default)

- Add DEFAULT_THRESHOLDS['drop'] = -5 and DEFAULT_HARD_DROP_RULES const
- Add evaluate_hard_drop(): fires when no positive signals + short latin name OR ≥3 placeholder fields
- Wire into classify(): drop overrides all other labels; synthetic reason explains trigger
- Add drop_reason_label() and 'drop' => '⛔ авто-дроп' to MessageFormatter::LABEL_MAP
- Update ClassifierTest/MessageFormatterTest to extend Cf7stgTestCase (filter isolation)
- Adjust pre-existing assertions that now correctly yield 'drop' instead of 'spam'

tests/ClassifierTest.php

@@ -4,9 +4,8 @@ declare(strict_types=1);
 namespace Cf7stg\Tests;
 
 use Cf7stg\Classifier;
-use PHPUnit\Framework\TestCase;
 
-final class ClassifierTest extends TestCase
+final class ClassifierTest extends Cf7stgTestCase
 {
     public function test_real_client_inquiry_is_green(): void
     {
@@ -23,14 +22,16 @@ final class ClassifierTest extends TestCase
 
     public function test_seo_spam_with_url_is_red(): void
     {
+        // score = SEO(-2) + URL(-3) + latin_only(-1) = -6, which is ≤ drop threshold(-5)
+        // → label promoted from spam to drop (score-based drop, no hard-rule since email is valid)
         $fields = [
             'your-name'    => 'Melissa Johnson',
             'your-email'   => 'promo@example.com',
             'your-message' => 'Hi, we boost SEO positions, visit http://cheap-seo.net',
         ];
         $r = Classifier::classify($fields, ['reason' => 'Akismet']);
-        self::assertSame('spam', $r['label']);
-        self::assertLessThanOrEqual(-2, $r['score']);
+        self::assertSame('drop', $r['label']);
+        self::assertLessThanOrEqual(-5, $r['score']);
     }
 
     public function test_honeypot_forces_spam_even_with_phone(): void