CommentRecorderParser sometimes generates ranges that miss Javadoc

The given test shows that CommentRecorderParser sometimes return incorrect ranges for elements as it may fail at including the Javadoc.
```java
// Regression test for Javadoc not being included in the declaration by
// CommentRecorderParser with all those preconditions
// 1. CU has a package
// 2. Javadoc above primary type
// 3. An annotation with arrayInitializer value
// 4. Some content to recover in the CU body
public void testRecoverCommentRecorderParserIncludesJavadocInDeclaration() {

CommentRecorderParser parser = new CommentRecorderParser(new ProblemReporter(
DefaultErrorHandlingPolicies.proceedWithAllProblems(),
new CompilerOptions(getDefaultJavaCoreOptions()),
new DefaultProblemFactory(Locale.getDefault())), false);
parser.setStatementsRecovery(true);
CompilationUnitDeclaration res = parser.parse(new org.eclipse.jdt.internal.compiler.batch.CompilationUnit("""
package p;
/** some javadoc */
@SomeAnnotationWithArrayInitializer({})
public class A {
B run= new B(
}
""".toCharArray(), "A.java", Charset.defaultCharset().toString()), new CompilationResult(new char[0], 0, 0, 0));
assertEquals(11, res.types[0].declarationSourceStart);
}
```