* Use std::array for IDNA status table

* Refactored punycode encoding

* Updated include guards for consistency

* Updated all header files

* Fixed regression in IPv4 address parser

* Reverted a change that is certainly a bug

* Fixed regression in string split function

* Cleaned up parser a bit more

* Refactored IP addresses

* Parsing ipv4 addresses now doesn't allocate dynamic memory

* Applied constexpr and inline across the library

* Added a string element range

* Changes to domain processing

* Refactored query parameter range

* Refactored query iterators

* Updated CMakeLists.txt

* Removed some redundant code

* Join takes a template argument instead of std::vector

* Fixed compilation error in ipv6 address

* Updated .appveyor.yml

* Updated search_parameters

* Fixed bug I introduced in the last commit

* Cleaned percent encoding code

* Configured filesystem dependencies

* Updated target link library

* Fixed bug in build scripts where the target_link_libraries were not set correctly for some tests

* Fixed regression caused in e0c32ad9

* Fixed second regression failure

* Updated internals of parser to avoid some copies of url_record

* Added a path iterator

* Inline all filesystem path functions

* Renamed a couple of headers

* Updated build scripts

* Cleaned up implementation of idna.cpp

* Rewrote documentation

* Updated remaining samples

* Added prototype for converting between URL components and JSON

* Added implementation and tests for converting a query string to a JSON data structure and back

* Added optional value to query_iterator

* Added error object for JSON query objects

* Moved some more deckchairs around

* Moved some more files around

* Added tests to convert from ascii encoded domains to unicode

* Refactored tests

* Renamed domain_to_unicode

* Added const accessor for search_parameters and changed return type of setter functions

* url::port<intT> now returns an optional<intT>

* Simplified error handling of percent encoding

* Applied trailing return type to functions in core and details

* Applied trailing return type wherever it made sense

* Added percent decoding to url_search_parameters

* Added test for url_search_parameters::append

* Moved query iterator into a new header, with rename

* Added function that simplies percent decoding URL components

* Added percent encoding function, updated spome public headers for API compatibility

* Moved a lot of files around

* Refactored internal string algorithms

* Added accessors to test if the hostname is a valid IP address or domain name

* Added tests for ipv4 address endianness

* Added missing header

* Added endianness header, needed for correct behaviour of ipv4_ and ipv6_address

* Added endianness.hpp to CMakeLists.txt

* Added check for hostname bounds in ipv6 address accessor

* Fixed warnings and errors on CI

* Added more unit tests for opaque hosts

Guessing from test names it was intended to test `https` and `file` schemas, not 3 duplicates for `http` schema