Because the net/http server removes \r\n from multi-line header values,
there's no need to to check for \r or \n when skipping whitespace in
headers (see https://godoc.org/net/textproto#Reader.ReadMIMEHeader).
Given this fact, the whitespace test can be simplified to b == ' ' || b
== '\t'. There's no need for the isSpaceOctet bit field in octetTypes.
The isTokenOctet bit field is the only bit field remaining after the
removal of isSpaceOctet. Simplify the code by replacing the
isTokenOctet bit test in octetTypes with an array of booleans called
isTokenOctet.
Declare isTokenOctet as a composite literal instead of constructing it
at runtime.
Add documentation to core functions for parsing HTTP headers.
merlin