More than 1 pod of unicorn+workhorse = unusable attachments
When we have more than one pod of unicorn, user uploads attachments that end up in one of the pods file system
In case the user refreshes the issue page the chance of seeing his attachment (being served by the exact workhorse that received it in the first place) is 1/n where n is the number of unicorn pods he has
cc: @twk3 @marin @WarheadsSE @joshlambert
We need to have shared persistent storage for these pods (also solves https://gitlab.com/charts/helm.gitlab.io/issues/307)