CommonGramsFilterFactory doesn't respect ignoreCase=true when default stopwords are used

[chenhh021]

See Lucene-10008. It's also valid for lucene.

CommonGramsFilterFactory's use of the "words" and "ignoreCase" config options is inconsistent with how StopFilterFactory uses them - leading to "ignoreCase=true" not being respected unless "words" is specified.

Reproduce:

[Test]
        public void testIgnoreCase()
        {
            IResourceLoader loader = new ClasspathResourceLoader(typeof(TestAnalyzers));
            CommonGramsFilterFactory factory =
                (CommonGramsFilterFactory)
                TokenFilterFactory("CommonGrams", TEST_VERSION_CURRENT, loader, "ignoreCase", "true");
            CharArraySet words = factory.CommonWords;
            assertTrue("words is null and it shouldn't be", words != null);
            assertTrue(words.contains("the")); //passes
            assertTrue(words.contains("The")); //fails
            Tokenizer tokenizer = new MockTokenizer(new StringReader("testing the factory"),MockTokenizer.WHITESPACE, false);
            TokenStream stream = factory.Create(tokenizer);
            AssertTokenStreamContents(
                stream, new String[] {"testing", "testing_The", "The", "The_factory", "factory"});
        }

Working for a PR now.

[chenhh021]

PR created. It doesn't port the whole Lucene patch where a new base class was created, but applies minimum change solve the problem.