[CodeQuality] Deprecate SwitchNegatedTernaryRector

[TomasVotruba]

Deprecates SwitchNegatedTernaryRector.

The rule flips ! $cond ? A : B into $cond ? B : A. The condition gets positive, but the branch values swap places — and that ordering often carried meaning.

Simple case is fine:

-return ! $upper
-    ? $name
-    : strtoupper($name);
+return $upper
+    ? strtoupper($name)
+    : $name;

Nested case is where it turns:

-return ! isset($result['validation_messages'])
-    ? []
-    : (isset($result['validation_messages']['http']) ? $result['validation_messages'] : $result);
+return isset($result['validation_messages'])
+    ? (isset($result['validation_messages']['http']) ? $result['validation_messages'] : $result)
+    : ([]);

The original reads guard-first: bail value [] up top, complex branch last. After the rule, the nested ternary is jammed into the ? branch and the trivial fallback sits at the bottom. Also note ([]) — parentheses added around a plain array literal.

Second fixture had the same artifact:

-return ! isset($a)
-    ? (isset($b) ? $b : 'b')
-    : $a;
+return isset($a)
+    ? ($a)
+    : (isset($b) ? $b : 'b');

($a) — parentheses around a bare variable.

Which form reads better depends on what the values mean, not on the shape of the condition. That is a personal preference, so the rule is deprecated rather than patched.

Changes:

• SwitchNegatedTernaryRector implements DeprecatedInterface, refactor() throws
• removed from CodeQualityLevel
• rule tests and tests/Issues/Issue6480 (which existed only to cover this rule's interplay with RemoveDeadInstanceOfRector) removed
• dropped the now-stale pointer to this rule from RemoveUselessTernaryRector