WL#3526: Incremental BASE64 encoding/decoding

# RATIONALE

The current set of functions for performing BASE64 encoding cannot handle
incremental encodings, i.e., encoding the data piecemeal. Instead you have to
accumulate the data into one big buffer in memory, and then encode it all in one
go. This works well for small chunks of data, but if huge amounts of data are to
be encoded, an incremental approach is necessary.

# PROPOSAL

Implement an incremental, stream-oriented, encoder/decoder that can write
to/read from an IO_CACHE.