Fix/concurrency

[youwe-petervanderwal]

Fix concurrency issue in QueueManager::processBatch

Error: Uncaught Amp\Beanstalk\DeadlineSoonException causing the ESB to crash

Caused by: Wrong Beanstalk job deleted at the end of the job
Caused by: Multiple Beanstalk jobs for the same job UUID. With multiple concurrent workers that same job UUID will be assigned to multiple workers. $uuidToBeanstalkIdMap[$job->getUuid()] will then only contain one of the beanstalk job ids, so one of the worker's queue manager won't be able to delete his job after finish.
Caused by: Concurrent calls to QueueManager::processBatch() were possible (and should be possible), but the batch itself was shared over multiple HTTP requests. Causing the same job being inserted twice (or more).

Resolved by: Processing the batch via a local variable instead of the shared property.

Steps to reproduce:

1. Create a flow with a HttpRequestProducer producing a job (or multiple jobs) on each call
2. Set the worker.instances config for that flow greater than one (I used 5 in my setup)
3. Fire some CONCURRENT!!! http calls to your ESB, for example with a bash command
   for n in {1..10}; do curl 'http://localhost:34981/my-url' 2>/dev/null >/dev/null & done
4. Wait for the workers to pick up the jobs ("Worker reserved a Job" debug messages)
5. Wait for the job TTR to expire (default $timeout variable in Job is 60 second)
6. In my setup I got a consistent DeadlineSoonException, but as always with concurrency errors that might only occasionally occur.

Ps. I cherry-picked some commits from my other PR's that were required for the travis-ci checks. The main commit in this PR is of course the first one.