Add replica-announced config option

[fatpat]

The 'sentinel replicas ' command will ignore replicas with
replica-announced set to no.

The goal of disabling the config setting replica-announced is to allow ghost
replicas. The replica is in the cluster, synchronize with its master, can be
promoted to master and is not exposed to sentinel clients. This way, it is
acting as a live backup or living ghost.

In addition, to prevent the replica to be promoted as master, set
replica-priority to 0.

The initial use case was the following:

we have a 3 nodes redis cluster, with 1 master and 2 replicas
we have a 3 nodes sentinel cluster
clients ask sentinel which redis is the master for writes
clients ask sentinel which redis are the replicas for reads
we have real time high load writes to the masters
we have real time high load reads from replicas
we need async, low frequency, heavy read load using complex and CPU consuming LUA scripts.
we want to ensure the async load will not impact performances on the replicas from which the LUA scripts are running
we want another replica to be replicated (#captainObvious) but we don't want it to be master nor to accept requests from normal clients
so we add replica-announced

This PR replace #8437

[fatpat]

@yossigo @hwware @itamarhaber

[yossigo]

@redis/core-team your approval is needed for this, as this adds a new configuration option and INFO field for Sentinel's use.

[oranagra]

LGTM (only looked at the new config and new info field).

My only concern was if we want to "pollute" redis with more features that are only used by sentinel at this point, but considering this is following the footsteps of replica-priority, i suppose that's fine.

[itamarhaber]

LGTM

[hwware]

just want to put some more thoughts of this feature, we can still aware the replica existance if we see "num-slaves" fields in sentinel masters/master command together with sentinel info-cache, the replica just not "offically announce" it. IMHO this shouldn't cause any issue, but bring benefit for debuging purposes. WDYT @yossigo @oranagra @itamarhaber

[madolson]

API LGTM

[yossigo]

@hwware I somehow missed your comment, sorry about that. I lean towards maintaining consistency between the number of reported replicas and the number, and if this proves to be a troubleshooting obstacle have a dedicated counter for these replicas. Hope this makes sense.

[yossigo]

Merged, thank you @fatpat!