[Java] Charset encoding handling improvements.

[vyazelenko]

This PR contains several improvements to how charset encoding is handled by the JavaGenerator:

• Add support for case-insensitive aliases for standard charsets, e.g. lAtIN1 will be resolved to StandardCharsets.ISO_8859_1. This is similar to how Charset.forName(String) works.
• Use StandardCharsets when generating static CharacterEncoding methods, e.g. java.nio.charset.StandardCharsets.ISO_8859_1.name() will be used instead of a string "latin1" (when characterEncoding=latin1).
• Use canonical names for charsets. The standard charsets are canonical by definition since the constants are used. For non-standard charsets the canonical name is resolved at the code generation time, e.g. if characterEncoding=cp912 is specified the resolved name will be ISO-8859-2 and will be used everywhere the original name would have been used. If the charset does not exist at the code generation time then the raw value is used.
• Use putStringWithoutLengthAscii for methods taking a CharSequence.
• Always use a Charset instance when converting to/from String. This eliminates most of the try/catch blocks in the generated code.
• Avoid creation of an empty byte[] when input string is null or empty.
• Treat character encodings as case-sensitive.