Streamline builder images

[ynohat]

With #42 we pinned all builder images to alpine 3.11, matching the BASE in dockerfiles/base.Dockerfile. This ensures
a matching distribution and interpreter runtime between builders and final images.

The problem is that it is hardcoded and duplicated across a few files, a minor maintainability issue.

Would it make sense to either:

• have a build arg to indicate which Alpine version we're using
  • implies that all builder base images should include the alpine version in the tag, e.g. golang:alpine3.11
• have one builder image base with all runtimes (node, goloang etc...), using akamai/base as BASE
• use akamai/base as the base for all builders, and install runtimes manually in each dockerfile

I think I like the "one builder image base" option, because it gives a lot of control, but I haven't thought through all the implications.