So, to be clear you're talking about merely contacting a majority of servers, but not necessarily verifying that the server that wants to become a candidate can be elected, right? Not to sound like a broken record, but doesn't the pre-vote protocol cover this? I mean, getting a majority of votes during the pre-vote phase also implies the server can reach a majority of members, and pre-vote will fail if either a majority can't be reached or if votes aren't received from a majority. So, if pre-vote covers this (since a majority must be contacted to receive a majority of votes) what's the reason for wanting to implement this but not pre-vote? You mentioned simplicity? Is it that much simpler than implementing the pre-vote protocol? Implementing the pre-vote protocol is pretty trivial. After all, presumably the logic for granting votes is already present in your implementation and should be reusable if it's designed well.
As for availability, I think if the pre-vote protocol implies "majority ping," and if the pre-vote protocol is assumed to be safe, you can assume "majority ping" is safe.