Redesign Streams and Character Encoding

[daumayr]

Right now Stream operations are very byte oriented and only ASCII characters are supported by the character converters. Streams should be redesigned to offer operations that read/write multiple bytes at once, making implementation of character encodings other than ASCII easier.
The Java implementation of strings and characters uses UTF-16 encoding, so this is going to be the encoding of the internal representation.

Goals:

• multi-byte operations (like nextInt nextLong nextDouble etc.)
• support reading/writing other Character Encodings
• more strictness / better Exceptions

[smarr]

First progress in that direction was made with #217.