Usage of --wait by default

[consideRatio]

I've come to think --wait is a very bad practice in Helm, and run into some issues right now while developing locally due to it.

A motivation is provided in consideRatio/consideratio.github.io#6, but in general the short version is that usage of --wait where someone choose not to wait the duration will put things into a bad state. Then, the value of --cleanup-on-fail can also be lost.

I think the only place --wait could make sense given its technical weaknesses, is in automation where one is always ready to await the full duration. Even there, it may be questionable though I think.

I'd love to see it removed by default somehow.

[yuvipanda]

The primary reason for '--wait' is that if it fails, we know the deploy failed. Otherwise you might end up in states - for example, when the hub image is wrong, or it is in crashloopbackoff, etc - where hubploy 'succeeds'.

Cancelling a deploy in the middle of a '--wait' definitely puts things in a weird state - a bug that keep hoping the helm people will fix. Ideally, this would be two commands - one to do the deploy, and one to wait for all resources to be ready. This should work around the issue.

I think if we can remove '--wait' while keeping it so that a hubploy deploy will fail if the deployment fails, I'd really really love that!