fix: Add precautionary logic for PDB

[krithika369]

What this PR does / why we need it:

PDB support was introduced in #420. In some scenarios, the PDB can create problems. As an example, if PDB is configured to have a minimum of 20% replicas available and there is only one replica in the deployment, it's not possible to drain that node. To prevent such issues, this PR introduces some precautionary checks, - ensuring that at least 1 replica of the deployment can be removed:

Let
===
# replicas: r
minAvailablePercent / 100: p

We want
=======
r - ceil(r * p) >= 1 // ceil because k8s rounds up to the nearest int
=> ceil(r * p) + 1 <= r
=> ceil(r * p) < r // We can say this because all terms in the above inequality are int

It's enough to check ceil(r * p) < r where r is the minReplicas because:

r - ceil(r * p) >= 1
=> r * (1 - ceil(p)) >= 1

We can see that, as r increases, LHS will increase. So, it's enough to ensure this inequality holds for the smallest possible value of r, which is the minReplicas configured for the deployment.

Does this PR introduce a user-facing change?:

Disable PDB when the min replicas is very low, so that it doesn't block node drain.

Checklist

• Added unit test, integration, and/or e2e tests
• Tested locally
• Updated documentation
• Update Swagger spec if the PR introduce API changes
• Regenerated Golang and Python client if the PR introduce API changes

[krithika369]

Removing createPodDisruptionBudgets from the controller so that it's easy to unit test it.

[ariefrahmansyah]

LGTM! Thanks for making it better, Krithika!