Fix implicit type‑conversion precision issues in equalsIgnoreCase

[Chenjp]

See testcase,

        // bc ByteChunk bytes: Hello
        Assert.assertTrue(bc.equalsIgnoreCase("heLLo"));
        Assert.assertFalse(bc.equalsIgnoreCase("\u8a48\u8a65\u8a6c\u8a6c\u8a6f"));

Ascii.toLower(c) implicitly assume c<=0xff.
It is hard to exploit with this weakness. Now add an explicit check to improve robustness.