[Cache] Don't store hop-by-hop headers

[krizhanovsky]

Keep-Alive and Server are hop-by-hop headers and should not be cached. hbh_hdrs must be adjusted and the headers must be made special to be analyzed quickly in tfw_cache_copy_resp().

Relates to #407.

[krizhanovsky]

See also What Proxies Must Do and particularly the referenced spec

[vankoven]

Actual problem differs from mentioned above: hop-by-hop headers are forwarded as end-to-end headers

In spec mentioned above 8 headers are marked as a hop-by-hop. But the final document RFC7230 is more strict than a draft and close to article What Proxies Must Do:

   A proxy or gateway MUST parse a received Connection header field before
   a message is forwarded and, for each connection-option in this field,
   remove any header field(s) from the message with the same name as the
   connection-option, and then remove the Connection header field itself
   (or replace it with the intermediary's own connection options for the
   forwarded message).

So, for request:

    GET /foo HTTP/1.1
    Host: example.net
    Connection: Keep-Alive, Foo, Bar
    Foo: abc
    Foo: def
    Keep-Alive: timeout=30

The Connection, Keep-Alive and both Foo headers must be removed before forwarding it.
Co-Advisor HTTP 1.1 compatibility tests follows obsoleted RFC2616, but also checks hop-by-hop headers removing from messages.

'tfw_http_set_hdr_connection()` seems to return without updating Connection header. Also other hop-by hop headers are not removed from test messages.

Used tempesta (af32a834) configuration:

server <test-server-ip>
cache_methods GET HEAD;
cache_fulfill * *;

[krizhanovsky]

Please also store correct Server and Via headers in cache as described in #634

[vankoven]

fixed in 72a24ae