Add gzip HTTP handler/client wrapper

[klauspost]

Fork and clean up+extend the dead nytimes/gziphandler project.

• Adds http.Transport wrapper.
• Includes Refactor out gzipwriter interface nytimes/gziphandler#106 as well as support for stateless encoding.
• Implements a variant of Add NotContentTypes option nytimes/gziphandler#81
• Fixes Incorrect check for short write nytimes/gziphandler#103
• Strip "Accept-Ranges" on compressed content. Fixes Range-Requests aren't properly handled nytimes/gziphandler#83
• Removes testify from deps.
• Constructors boiled down.
• Defaults to this gzip package.
• Allocations reduced.

Default settings comparison:

λ benchcmp before.txt after.txt                                        
benchmark                         old ns/op     new ns/op     delta    
BenchmarkGzipHandler_S2k-32       51302         25554         -50.19%  
BenchmarkGzipHandler_S20k-32      301426        174900        -41.98%  
BenchmarkGzipHandler_S100k-32     1546203       912349        -40.99%  
BenchmarkGzipHandler_P2k-32       3973          2116          -46.74%  
BenchmarkGzipHandler_P20k-32      20319         12237         -39.78%  
BenchmarkGzipHandler_P100k-32     96079         57348         -40.31%  
                                                                       
benchmark                         old MB/s     new MB/s     speedup    
BenchmarkGzipHandler_S2k-32       39.92        80.14        2.01x      
BenchmarkGzipHandler_S20k-32      67.94        117.10       1.72x      
BenchmarkGzipHandler_S100k-32     66.23        112.24       1.69x      
BenchmarkGzipHandler_P2k-32       515.44       967.76       1.88x      
BenchmarkGzipHandler_P20k-32      1007.92      1673.55      1.66x      
BenchmarkGzipHandler_P100k-32     1065.79      1785.58      1.68x      
                                                                       
benchmark                         old allocs     new allocs     delta  
BenchmarkGzipHandler_S2k-32       22             19             -13.64%
BenchmarkGzipHandler_S20k-32      25             22             -12.00%
BenchmarkGzipHandler_S100k-32     28             25             -10.71%
BenchmarkGzipHandler_P2k-32       22             19             -13.64%
BenchmarkGzipHandler_P20k-32      25             22             -12.00%
BenchmarkGzipHandler_P100k-32     27             24             -11.11%

Client Transport:

Speed compared to standard library for an approximate 127KB payload:

BenchmarkTransport

Single core:
BenchmarkTransport/gzhttp-32         	    1995	    609791 ns/op	 214.14 MB/s	   10129 B/op	      73 allocs/op
BenchmarkTransport/stdlib-32         	    1567	    772161 ns/op	 169.11 MB/s	   53950 B/op	      99 allocs/op

Multi Core:
BenchmarkTransport/gzhttp-par-32     	   29113	     36802 ns/op	3548.27 MB/s	   11061 B/op	      73 allocs/op
BenchmarkTransport/stdlib-par-32     	   16114	     66442 ns/op	1965.38 MB/s	   54971 B/op	      99 allocs/op

This includes both serving the http request, parsing requests and decompressing.