Path segment normalization should be consistent with RFC 3986 Section 5.2.4

[sethmlarson]

https://tools.ietf.org/html/rfc3986#section-5.2.4

[sigmavirus24]

Can you add tests first that are based in the RFC that show we're not consistent with it?