* #630 only set status code if response hasnt started, otherwise exception
* #623 made {RemoteIpAddress} available as placeholder so you can do x-forwarded-for
* #623 local address different on mac, windows and linux for integration test
* renamed some middlewares so they make more sense
* add downstreamroutefindermiddleware benchmark
* #574 consolidate some code, man the config stuff is a mess!
* #574 just use the downstream re route and the key for caching http clients
* #574 added benchmark, i was suprised to learn using a complex type was faster than a string in benchmark .net dictionary tests, hey ho probably dont have enough data in the type...
* #464 added code to request mapper to not automatically add content type and content length headers, .net will automatically try and add these headers in a few circumstances but this solves the 464 issue
* #464 use seek instead of read on body check for websockets tests
* #464 ran out of inodes on linux, looks like reloadonchange causes this
* removed file
* updated package
* updated package
* updated package
* updated package
* updated package
* updated package
* updated package
* all packages updated
* #295 can add token to service provider config and this will be used by consul clients to get services and configuration
* #295 wait longer for this test
* started messing around with this on the train last night
* mega hacking away to change middleware into Ocelot iddleware
* scoped data repo back in
* broken commit getting tests working
* another broken commit farting around with tests
* all unit tests passing again
* mw pipeline for ocelot...still loads of hacks but getting there now to get acceptance tests working, then fix config so you can have aggregate and then imlement multiplexer, then mapping to response...loads to do
* all tests passing before aggregation feature implemented
* removed all the request middleware stuff we dont need it
* updated how errors work...tho i think there could be edge case here when aggregating because one downstream could error and this would effect another
* removed multiplexer so you dont have to send route down, this isnt very thread safe...sigh
* hacking around getting the config for aggregates in, this might change
* refactored builder and unit tests passing now
* Updated a bunch of ports for tests
* plugged in code to create reroutes that are aggregates
* made multiplexer a class
* hacked test to death
* simple aggregator done, initial validation done
* removed request id from context, it is still specific for http request
* now aggregates to json always
* docs for aggregate reroutes
* Updated docs
* hacked together load balancing reroutes in fileconfig
* some renaming and refactoring
* more renames
* hacked away the old config json
* test for issue 213
* renamed key
* dont share ports
* oops
* updated docs
* mvoed docs around
* port being used
