gpexpand : Fix multi port issue when clusters are not in balanced state

[shrakesh]

Why : In previous PR 13757 where the plan was to check the health of the cluster before starting the expansion in gpexpand. while backporting in 6x we realised that hard exit after the health check will create problem for the customer when due to some reason they will not able to make the cluster up in balance state. this will block customer to expand.

Issue : we found the route cause as when the cluster is not in balance state then the max/min port is returning wrong port values and as result system is expanding at the wrong port which is giving multi port error.

Fix: updated the function to work in imbalance cluster state scenario. we are collecting all the port for primary and mirror based on preferred role of the segment, and if the role is switched we are considering segmentPairs.mirrorDB for primary port and segmentPairs.primaryDB for mirror port. in this we are collecting the port which is based on the preferred role.

another change in the pr is, instead of error exit in imbalance scenario we are just logging the warning.

Added unit test cases for the positive/negative and exception scenario for the newly added functions. Updated feature test case where we are converting error to warning.

Here are some reminders before you submit the pull request

• [ ] Add tests for the change
• [ ] Document changes
• [ ] Communicate in the mailing list if needed
• [ ] Pass make installcheck
• [ ] Review a PR in return to support the community