KAFKA-12158: Better return type of RaftClient.scheduleAppend

[dengziming]

More detailed description of your change
It's not very convenient to return Long for RaftClient.scheduleAppend, in this PR we add a RaftAppendResult class to replace it.
I also found that we lack logic on append failure, for example, the ControllerWriteEvent won't check the result of raftClient.scheduleAppend, this should be fixed in a future PR.

Summary of testing strategy (including rationale)
QA

[hachikuji]

cc @jsancio

[jsancio]

Thanks for the changes @dengziming

[hachikuji]

Thanks for the updates. Left a few suggestions.

[jsancio]

Thanks for the chances, just a few minor suggestions.

[hachikuji]

I think it would be nice if we could consolidate NotLeaderException and FencedEpochException. In the context of scheduleAppend, they mean the same thing to the caller. They both say that the raft state has transitioned and the operation is no longer possible. I'm inclined to keep NotLeaderException and document that it covers both cases. What do you think?

[dengziming]

Here I referred to the practices of KafkaController in which we have NOT_CONTROLLER and STALE_CONTROLLER_EPOCH for RPC errors, but here I also think it's better to consolidate NotLeaderException and FencedEpochException now.

[jsancio]

One observation, otherwise LGTM.

[jsancio]

@dengziming @hachikuji What do you both think about making this consistent with RaftClient::resign? RaftClient::resign throws IllegalArgumentException when the passed epoch is greater than the current epoch. With this change KafkaRaftClient::schedule*Append throws NotLeaderException when the passed epoch is greater than the current epoch.

Maybe something like:

if (epoch > this.epoch) {
    throw new IllegalArgumentExcpetion("...");
} else if (epoch < this.epoch) {
    throw new NotLeaderException("...");
}

[dengziming]

I think this is a good suggestion, only an epoch smaller than the current leader epoch should be treated as NotLeaderException.

[hachikuji]

LGTM. Pushed one trivial change in ReplicatedCounter.