the e638f9fb introduced deduplication
for queries over UDP, the idea is to track same queries and answer
only once, however that introduces both constant processing and
memory overhead and could break clients who count packet loss
by queries sent/received. disabling for now.

* in the begin() layer, the incoming query is
  exposed as req->qsource.packet, it is invalidated
  after begin() and should not be modified
* the destination address (local interface) is
  also tracked for filtering purposes

previously the buffer for TCP pkt reassembly was
not correctly cleared and fragmented answers
were rejected 

* when enqueued task terminated earlier than leader
  task because of timeout, it wasn't dequeued from
  the waitlist immediately, but it didn't have any
  outstanding outbound queries. when leader task
  terminated, it removed this task and updated its
  outbound query, which didn't exist. this triggered
  a 16B write in undefined location
* fixed timeout timer being scheduled for closing
  without holding reference to parent task

many clients do frequent retransmits of the query
to avoid network losses and get better service,
but then fail to work properly when a resolver
answers SERVFAIL to some of them because of the
time limit and some of them NOERROR. 
it's also a good idea to avoid wasting time
tracking pending tasks to solve the same thing.

if the upstream TCP query timeouted or the connection
was severed, it would dissociate the handle from
original query, so the query would be solved
but the requestor wouldn't see the answer unless
he requeried

* daemon now processes messages over TCP stream
out-of-order and concurrently
* support for TCP_DEFER_ACCEPT
* support for TCP Fast-Open
* there are now deadlines for TCP for idle/slow
streams (to prevent slowloris; pruning)
* there is now per-request limit on timeouts
(each request is allowed 4 timeouts before bailing)
* faster request closing, unified retry/timeout timers
* rare race condition in timer closing fixed

resolver will always attempt to contact upstreams
known to be bad if it's not busy.
this fixes a problem on low-volume resolvers
where a short connection outage could make
resolvers deny resolving queries even after the
connection is restored

previously full timeout led to reset of the evaluated
address list and no upstream server was penalised
for not answering the query, this penalises all of
tried servers with TIMEOUT

lib: answer finalization was changed

this is not going to be backwards compatible change, but it will be the first tagged libknot release sufficient for resolver

* PIE,RELRO+NOW and other security features enabled
* support for both static/dynamic builds with BUILDMODE
* dynamic library is ABI-versioned, starting at 1
* pkg-config file is installed

worker can track outbound requests and if N resolutions want the same
subrequest, only one will lead it and others will be notified when it
finishes

this massively reduces number of outbound requests for
slow/unresponsive/low ttl requests

this also fails if the timeout timer cannot be started,
as it would wait for undefined time otherwise

instead of single I/O request per step, the daemon now retries
all addresses in the selection with 300ms timeout between tries.
there are len(list) + len(list)/2 tries

the idea is to reduce latency when UDP request doesn't punch through,
or some NSs are overwhelmed/faulty

if the library or daemon is compiled with address sanitizer, objects in
freelists are poisoned to detect use-after-recycle errors
it is not currently used in the library, as there are no freelists

this is going to be needed for issuing multiple timed queries + fast
retry

the selected address is now an array with selection, caller can then send the same query to multiple offered targets

refs #35