This introduces diffx.reader.DiffXReader, a streaming parser for the

DiffX file format. This is a low-level interface for DiffX files, which

is able to read from a file stream (such as a local file, HTTP response,

or memory-backed stream) and parse and return each section in the DiffX

files according to the specification.
The parser acts as a generator, and will provide dictionaries containing

information on each section of the DiffX file. This includes options,

metadata or text/diff content, the section ID, section type, and section

hierarchy level. Consumers can build upon this to work with the data as

it comes in, or to easily convert it into another representation.
As a streaming parser, this does not keep much state around between

sections. This makes it ideal for working with very large DiffX files,

or reading from a stream that may incrementally provide new content.

However, it also means that a parsing error may not manifest until later

in the stream, after the consumer has already handled some content. An

object-based implementation is in the works that will address this.
As this is a reference implementation, the parser is very strict about

conforming to the specification, when it comes to header structure,

characters in header options, lengths, and valid option values. It does

take into consideration different newline formats and ignores extra

newlines before/after sections (including at the beginning or end of the

file).
At this stage, DiffXReader should be usable for any production use of

the DiffX format. An accompanying DiffXWriter, and implementations

representing DiffX through an object model, are in the works.

Unit tests pass on Python 2 and 3.
Put the reader through its paces by testing a bunch of valid and invalid

DiffX files generated by hand and by the upcoming DiffXWriter.