Reduce broken TravisCI builds by using ipv4-only sks-keyserver pool

[clingle]

When submitting my previous PR I noticed that both the current master branch and my PR CI build failed due to the gpg keyserver address not being found, and found this similar issue.

TL;DR Docker containers don't generally have access to IPv6, sometimes ha.pool.sks-keyservers.net gives IPv6 addresses to use for GPG. Changing the pool to only return IPv4 addresses should fix this.

[clingle]

I just realized I didn't run update.sh after making the change to the template files (doh!). I'm not sure what the preferred action is here, but I can either:

1. close this PR and resubmit with the correct files
2. submit an additional PR with just the files modified from running the update.sh script

Pending the result of PR #601 it may be better to wait on this one until it's decided whether the two additional Dockerfiles in that will be added to master or not.

[tianon]

In our experience across all the Docker official images, the ipv4 pool isn't any more reliable on the whole than the ha pool (and in fact is made up of a number of the same servers). I think you probably just happened to hit a different server via ipv4 than you did via ha.

[clingle]

That may well be the case, so I have no issue in closing this if it's deemed unnecessary from your experience.