RetryUntilMajorityCommit::untilStepdownOrSuccess is the intended API, but it is currently limited to callers that have a CancelState object. The resharding coordinator uses CoordinatorCancellationTokenHolder (_ctHolder), so introducing a separate CancelState would require maintaining two parallel cancellation abstractions for the same coordinator instance. That makes RetryUntilMajorityCommit difficult to use in this context and led to the introduction of a separate resharding::runUntilSuccessOrStepdown helper.

We should either extend RetryUntilMajorityCommit so resharding can use it without a separate CancelState (for example, via an adapter from _ctHolder or a token-based constructor), or refactor resharding's _ctHolder so it can use the existing helper directly.