Base64 decoder tried to handle several different cases in one fn
including how the data should be decoded and corresponding consumed and
decoded bytes calculated for RFC 2045 and RFC 4648 which are very
different. This left window for mistakes and edge cases that one mode is
not supposed to take care of while the other can.
Separate the decoder per RFC so as to have each variable their own scope
and playground. This also makes it much easier to debug if there are any
issues in decoding w.r.t. different callers.
This also takes care of some known issues like making sure the consumed
bytes are correctly handled.