[fix] always link zlib static

Hey @bomgar @emk

I'll take care of encoding @bomgar 's test and travis config with cargo new, sed and echo semantics today.

imho this project is insanely useful for rust's ecosystem and accessibility, so if this is the price to pay to get this merged and improve the project, I'm glad to oblige. Your project your rules, yadda yadda.

That said, here's the things that are bugging me about this @emk - you don't need to respond to any of that, but I hope this helps clear things up why your request is seing so much pushback from @bomgar . These observation might or might not be accurate, it's just what I'm thinking.

Saying this will cost you maintenance time is really backhanded, I mean currently there's no CI at all. So even if you end up spending 5 minutes this year on fixing a -v issue, how is that not an improvement to right now (where you have to run the test script after every change?)
Your woes with -v are basically that something else broke backcompat, right? I mean travis changing the UID is travis breaking backwards compatibility with regards to the UID. I'd argue that if something in the call chain is breaking backcompat, you'll always end up having to fix something up, and this has nothing to do with the docker volume feature. If docker does backwards incompatible changes to their CLI interface, you'll have to fix it too. Is that a reason not to use the CLI? Same reasoning about cargo new changing or the Cargo.toml format changing. Granted the UID thing is a bit more likely to go awry than cargo changing the cli interface, but not by much. I can see you reasoning being that it's better to reduce the surface that can break (eG don't use volumes if not necessary) and I understand that, but I don't think it's worth the tradeoff in that cases (losing simplicity and readability).
I've had UID issues before too, but honestly this has never taken more than a few minutes to fix. On a throwaway system like travis you can also simply chmod 777 and never think about it again. It's simply good enough
I've often seen people who were "afraid" to use tools or features because it bit them in the past. Like people being afraid to use git because they lost some work in a rebase and didn't know about reflog. I don't know and can't know if this is such a case, but this is very unhealthy imho.
You seem to assume that @bomgar will just get his changes in and never be seen again, why not ask "if the -v thing breaks will you fix it?", that seems like a win/win

That said these are just my thoughts. I'll stop that and take care of the test now ;)

emk / rust-musl-builder

[fix] always link zlib static #41