cast char to unsigned before isupper/islower in xpath and pattern lexers

[abdul7056]

Two lexing helpers in the C++ runtime hand a raw char to isupper()/islower():

• ParseTreePatternMatcher::tokenize branches on isupper/islower of tagChunk.getTag()[0]
• the XPath ID lexer action does the same on getText()[0] (source in XPathLexer.g4, kept in sync)
• a tag or xpath identifier whose first byte is non-ASCII (a UTF-8 lead byte such as 0xC3) makes that char negative, which is undefined for the ctype functions since the argument must be representable as unsigned char or EOF

Cast each byte to unsigned char before the call, matching the Java runtime's Character.isUpperCase/isLowerCase path.