Zookeeper jobstore support

[jvillar]

I'm using Zookeeper for HA coordination of distributed systems and I've implemented a job store for Apscheduler that uses Zookeeper as storage backend.

I've implemented tests and everything seems Ok, with a coverage of 98% for python 2.7 and python 3.4 and all tests passed.

Tox output:

apscheduler/jobstores/zookeeper.py 118 2 98% 127-128
py27: commands succeeded
py34: commands succeeded

[agronholm]

Thanks, looks very professional work! I will merge this soonish. Meanwhile, please check why the Travis run failed.

[coveralls]

Coverage increased (+0.2%) to 93.461% when pulling 3074469 on jvillar:master into 596c679 on agronholm:master.

[coveralls]

Coverage increased (+0.3%) to 93.508% when pulling 829777a on jvillar:master into 596c679 on agronholm:master.

[coveralls]

Coverage increased (+0.3%) to 93.508% when pulling 77a1e91 on jvillar:master into 596c679 on agronholm:master.

[coveralls]

Coverage increased (+0.3%) to 93.508% when pulling 19d8427 on jvillar:master into 596c679 on agronholm:master.

[jvillar]

Hi!
the problem has already been fixed. When asking for the children of a zookeeper node, Kazoo does not guarantee any order but it was returning an ordered list on my local tests. The order of the entries is now explicitly defined by the zookeeper jobstore class.