Do not check the pod's deletionTimestamp when determining if a pod is active

[rifelpet]

Fixing a regression originating from #173, this was causing nil entries in the pod index which resulted in credentials failing to be issued.

See #178 for more information.

[coveralls]

Coverage increased (+0.2%) to 20.0% when pulling 49a5e5d on rifelpet:deletionTimestamp into 80efbb1 on jtblin:master.

[lbernail]

Thanks a lot!

[timm088]

@jtblin how do you view this one in current release plan?
Looks to be a fix that keeps #173 related changes around pods 'missing' credentials during terminate

[rifelpet]

@jrnt30 was involved in merging and releasing #173 - any chance you can do the same for this?

[lbernail]

Quick follow-up: We've been running this for about 3 weeks without any issue and everything works fine

[jrnt30]

Thanks for the updates. Sorry for taking so long to merge this.

As mentioned previously I am not actively using kube2iam and was unable to consistently reproduce the issue when I had tried before and lost track of this.

Based upon the community feedback that this is working properly going to merge in. Going forward for PRs I will try and get an RC release process in place so we can "move the project forward" but also allow a bake in period for people wanting to use the latest updates in their own environment without touching ":latest"