Helm: Support minAvailable on connect injector PDB

[hamishforbes]

It looks like the PDB for the connect injector has copied the logic from the Server PDB.
In that it is extremely conservative and attempts to maintain n+1 pods, which makes complete sense for Consul Servers as you do want to maintain a quorum at all times.

The injector deployment uses leader election and only 1 pod is actually active at any one time, requiring n/2-1 pods to be available is unnecessary.

Furthermore, with the default 2 replicas maxUnavailable is set at 0, which requires manual human operator intervention to, for example, drain a node running an injector pod.
See Single-instance Stateful Application

For the connect injector deployment a minAvailable configuration makes much more sense.

These changes do not change the default behaviour, although I think the default behaviour probably should be changed.
This is opt-in only, users must configure the new value explicitly.

Changes proposed in this PR:

• Add connectInject.disruptionBudget.minAvailable to values
• minAvailable is null by default
• minAvailable takes precedence over the maxUnavailable configuration

How I've tested this PR:

• Added tests

How I expect reviewers to test this PR:

Checklist:

• Tests added
• CHANGELOG entry added

  HashiCorp engineers only, community PRs should not add a changelog entry.
  Entries should use present tense (e.g. Add support for...)

[kschoche]

Thanks for this PR!