Follow-ups from GlobalSign UTF8STRING issues

Goals:

[ ] Make sure that when a warning to GlobalSign is triggered, we see it in the daily report. How to evaluate: Read a daily report and see at least 1 warning logged as part of requests from asheeshdev.sandcats.io. Doable as soon as May 20 evening.
[ ] Make sure the sandstorm code change to switch to UTF8STRING is fully deployed. How to evaluate: Verify that there are 0 warnings (aka "success with errors logged") in a daily report. Doable as soon as May 24 evening.

Other reliability notes:

Sometimes GlobalSign's WSDL is down. Should we do something about that? Retry?
Sometimes (I think) if we fail to get MSSLDomainInfo due to a transient GlobalSign API failure, then we add something invalid to the cache rather than re-fetch MSSLDomainInfo.

Other engineering notes on the incident:

I don't have a machine anywhere that uses the GlobalSign dev API (but also maybe it doesn't have this issue?).
Maybe when we get XML parse errors in our SOAP responses, it's a CloudFlare page. If so, a retry would be OK, probably.

sandstorm-io / sandcats