Add parsing support for Windows

[chipsenkbeil]

In order to get some of our optimizations, we used memchr and variants to find linefeed characters (\n). This results in carriage returns (\r) not being captured to denote a newline such as with this example:

== some header ==\r\n
some paragraph of text

The header will fail as content parsed is == some header ==\r.

We need to go through our utility parsers and update all uses of the memchr to also cover carriage returns by checking for either \r or \n and, if finding \r checking to see if it is immediately followed by \n.

[chipsenkbeil]

Easiest way would be to remove the checks for '\n' and instead use nom's newline function to scan for it. Should see what the slowdown is using timekeeper.