Do not reuse the same timeout during connection timeout

[enidgjoleka]

Hi,

This PR tries to address two points:

• kpro_connection:start/3 spawns a new process which calls init/4. This fun was using the same connection_timeout during different phases of establishing the connection.
  This PR tries to use the "remaining timeout" in the different phases.
• kpro_brokers:discover_partition_leader/4 was calling kpro_lib:ok_pipe/1 without a timeout. It is true that now there is only one function that does network calls, kpro_connection:request_sync/3 but better be safe.

Please let me know if I missed something!

[enidgjoleka]

There is a failing test case which I am not sure is related to my change:

in call from kpro_txn_tests:test_txn_produce_2/2 (/home/travis/build/klarna/kafka_protocol/test/kpro_txn_tests.erl, line 164)
**error:{badmatch,{error,[{<<"test-topic">>,0,concurrent_transactions}]}}
  output:<<"">>

[enidgjoleka]

All the tests pass in both in OTP-21/OTP-22 in my local machine and I can not seem to reproduce this test failure.

It would be nice if I could get some traction on this PR.

[CLAassistant]

All committers have signed the CLA.

[zmstone]

why isn't this merged yet.

[enidgjoleka]

@zmstone to be honest I don't know, maybe because there is a failing test case.

[robertoaloi]

I think @enidgjoleka is correct:

module 'kpro_txn_tests'
  kpro_txn_tests: txn_produce_test...[2.297 s] ok
  kpro_txn_tests: txn_fetch_produce_test...[0.126 s] ok
  kpro_txn_tests: txn_produce_2_tx_test...*failed*
in function kpro_txn_tests:'-test_txn_produce_2/2-fun-2-'/6 (/home/travis/build/klarna/kafka_protocol/test/kpro_txn_tests.erl, line 155)
in call from kpro_txn_tests:test_txn_produce_2/2 (/home/travis/build/klarna/kafka_protocol/test/kpro_txn_tests.erl, line 164)
in call from eunit_test:'-mf_wrapper/2-fun-0-'/2 (eunit_test.erl, line 273)
in call from eunit_test:run_testfun/1 (eunit_test.erl, line 71)
in call from eunit_proc:run_test/1 (eunit_proc.erl, line 510)
in call from eunit_proc:with_timeout/3 (eunit_proc.erl, line 335)
in call from eunit_proc:handle_test/2 (eunit_proc.erl, line 493)
in call from eunit_proc:tests_inorder/3 (eunit_proc.erl, line 435)
**error:{badmatch,{error,[{<<"test-topic">>,0,concurrent_transactions}]}}
  output:<<"">>
  [done in 2.540 s]
=======================================================
  Failed: 1.  Skipped: 0.  Passed: 55.
===> Error running tests
Makefile:39: recipe for target 'eunit' failed
make: *** [eunit] Error 1

[zmstone]

likely caused by inter test case race condition.
will follow up with another issue which should not be a blocker for this PR.