Use Elasticearch readiness port instead of / for Pod readiness

[pebrc]

elastic/elasticsearch#50187 introduces a new readiness TCP endpoint we should use to determine pod readiness as it will take cluster membership into account https://www.elastic.co/guide/en/elasticsearch/reference/current/advanced-configuration.html#readiness-tcp-port

Our current readiness probe does call / which returns 200 before the node has joined a cluster. When rotating master nodes in quick succession through external means (so not through the rolling upgrade the ECK opererator orchestrates) for example a k8s cluster upgrade or cluster-autoscaler activity, the cluster might become unavailable for a few seconds because of that.

[pebrc]

The readiness feature is available as of Elastic stack 8.2.0. The implementation should include an appropriate version gate

[pebrc]

Njal foresaw this issue way back in 2019 #916