Small optimizations for the base64 decoder

[ahrav]

Additional optimizations to the base64 decoder.:

• Remove the bytes.Contains and replace with a constant time lookup using a b64 charset mapping using a fixed array.
• Instead of using a bytes.Buffer to store the characters of a substring, track the start and end positions of a substring in data during getSubstringsOfCharacterSet.
• Handle chunk Data replacement in a single pass.

[dustin-decker]

Nice! One comment on a potential improvement.