Improve INFO command for Sentinel internal failover feature

Redis' Info commands return a lot of information, but only these are required by Sentinel:
For slave:

run_id:4ba98afd1b8655725d9cbe64d9551bde0ceeffa0
role:slave
master_host:127.0.0.1
master_port:6381
master_link_status:down
slave_repl_offset:0
master_link_down_since_seconds:-1
slave_priority:100
slave_read_only:1
replica_announced:1
connected_slaves:0
master_failover_state:no-failover

Master

run_id:5a4dfcf77726cfcda754cbdc0a42d1cec6a7348d
role:master
connected_slaves:1
slave0:ip=127.0.0.1,port=6379,state=online,offset=42,lag=0
master_failover_state:no-failover

Explain:

• master_link_down_since_seconds: now() - last successful ping to master (or last ping received from master)
• master_failover_state: TBU
• master_link_status: current slave cannot contact with master
• slave_read_only: slave only executes read request
• slave_priority: a criteria used when promoting slaves to be new master (TBU)
• slave_repl_offset: TBU
• run_id: unique id of an instance, only null on the first run
• slavex:ip=....,port=...,state=...,offset=...,lag=... information of master about slave x, there can be multiple line like this, but the index of a slave should be consistent between calls
  Things to consider:
• Should we still use text based protocol and read line by line
• Should these information be displayed in external info command exposed to user's client, or keep it as an internal command

The issue names no files or tests. Begin by tracing the INFO command and Sentinel failover handling, then resolve whether the text protocol and external INFO exposure should change; completion requires defining the required fields and their semantics.