[helm] Add maxUnavailable option for rolling updates

[hwoarang]

Allow user to set maxUnavailable option for rolling updates in order
to speed up deployment times on large clusters.

[coveralls]

Coverage remained the same at 19.481% when pulling 56ee37c on hwoarang:add-ds-update-strategy into c24c3c8 on jtblin:master.

[jtblin]

cc @mariusv @walkafwalka can you help reviewing helm charts changes pls?

[mariusv]

LGTM 👍

[mariusv]

Please add documentation about the maxUnavailable on the readme on how it will be used.

[walkafwalka]

I prefer using something more flexible for updateStrategy but that would require a breaking change. We should take note of situations like this.

  updateStrategy: {}
  #  rollingUpdate:
  #    maxUnavailable: 1
  #  type: RollingUpdate

Besides the documentation, LGTM.

[rdelpret]

@walkafwalka I 100% agree, I reached out to @mariusv on slack about it to propose the same before I saw this PR but didn't get an answer.

[rdelpret]

Regardless of implementation, this one is urgent for us as large clusters need some pretty high timeouts set w/ helm to get through a rolling update with the hard-coded settings.

[hwoarang]

I have updated the default value and added the missing doc. I opted not to implement the braking change at this point. However, if you think we should do it then I can do that no problem