Remove individual polling behaviors in healthcheck

[epixa]

At the moment, each step of the healthcheck has its own polling behavior, but the healthcheck itself is also polled, so there does seem to be some unnecessary complexity here. I think the healthcheck would essentially behave the same way if we removed the individual polling and just relied on the global polling of the healthcheck.

One difference in behavior with this approach would be that every healthcheck would run every request on each poll, even when some sort of failure case was handled on the previous poll. The current behavior is to repeat only the current action until a success state for that action is encountered. It's unclear to me whether that's a problem or not, and it may even be more appropriate to repeat the whole series of checks each time anyway.

[jbudz]

We're down to one, validating elasticsearch's version. I think we're good on this one for now.

Related #32428
Related #33579