test: refactor test-http-destroyed-socket-write2

[santigimeno]

Remove the limit of requests to be sent (128) as in some conditions it
was reached without the error event being fired, causing the
test to fail.
Remove the initial timeout.
Remove some variables used to check the validity of the test and replace
them with common.mustCall and common.fail calls.

The error I was getting from time to time in OS X was:

=== release test-http-destroyed-socket-write2 ===                    
Path: parallel/test-http-destroyed-socket-write2
assert.js:89
  throw new assert.AssertionError({
  ^
AssertionError: false == true
    at test (/Users/sgimeno/node/node/test/parallel/test-http-destroyed-socket-write2.js:89:5)
    at Immediate.write [as _onImmediate] (/Users/sgimeno/node/node/test/parallel/test-http-destroyed-socket-write2.js:29:7)
    at processImmediate [as _immediateCallback] (timers.js:392:17)
Command: out/Release/node /Users/sgimeno/node/node/test/parallel/test-http-destroyed-socket-write2.js

[r-52]

CI https://ci.nodejs.org/job/node-test-pull-request/1444/

[Trott]

Looks like there's a minor lint issue to fix:

./node tools/eslint/bin/eslint.js src lib test tools/eslint-rules \
    --rulesdir tools/eslint-rules --quiet

/usr/home/iojs/build/workspace/node-test-linter/test/parallel/test-http-destroyed-socket-write2.js
  44:1  error  Trailing spaces not allowed  no-trailing-spaces

��� 1 problem (1 error, 0 warnings)

[santigimeno]

Ouch, sorry about that. Updated.

[Trott]

CI again: https://ci.nodejs.org/job/node-test-pull-request/1445/

[Trott]

Because of its history of flakiness, it might be a good idea to run a stress test for this on AIX and Windows.

It shouldn't stop this from landing, but it might also be worth experimenting after this lands to see if EPIPE and/or ECONNABORTED can be removed. If not, maybe the test can be modified to run until it receives an ECONNRESET since that's what the test was written to check for.

[Trott]

I don't think we have AIX in CI yet.

I kicked off a stress test on Windows 10. https://ci.nodejs.org/job/node-stress-single-test/395/nodes=win10/

[jasnell]

LGTM if CI is green.

[r-52]

LGTM

[Trott]

@santigimeno I'd like to land this, but I want to make sure we have a pretty recent CI run so I'd like to run it again. Can you rebase against master and force push?

[santigimeno]

@Trott done

[Trott]

CI: https://ci.nodejs.org/job/node-test-pull-request/1705/

[Trott]

CI is green. LGTM.

Would be interested to know if this fixes the issue for @thealphanerd too.

[Trott]

Landed in b920d45