Fix slowlog crash when redirection fails

[doyoubi]

Corvus may crash when it fails to connect to redis and try to create a sub-slowlog for multiple key command.

The Buggy Code

struct slowlog_entry *slowlog_create_sub_entry(struct command *cmd, int64_t total_latency)
{
    ...
    int64_t max_remote_latency = 0;
    struct command *c, *slowest_sub_cmd = NULL;
    STAILQ_FOREACH(c, &cmd->sub_cmds, sub_cmd_next) {
        // When this command fails because of not being able to connect to redis,
        // `remote_latency` will be zero (or maybe nagative) which result in a NULL `slowest_sub_cmd`.
        int64_t remote_latency = c->rep_time[1] - c->rep_time[0];
        if (remote_latency > max_remote_latency) {
            max_remote_latency = remote_latency;
            slowest_sub_cmd = c;
        }
    }
   // NULL `slowest_sub_cmd` will create a crash here.
    return slowlog_create_entry(slowest_sub_cmd, max_remote_latency / 1000, total_latency);
}

How To Reproduce

(1) Use the following Python 3 script to mock a redis cluster and let corvus connect to it.
Note that the slowlog-log-slower-than should be 0 and slowlog-max-len should be positive.

import time
import socket

server_address = ('localhost', 10000)
sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
sock.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
sock.bind(server_address)
sock.listen(1)

CLUSTER_NODES = b'*2\r\n$7\r\nCLUSTER\r\n$5\r\nNODES\r\n'
CLUSTER_NODES_RESP = 'ffffffffffffffffffff00000000490000000265 127.0.0.1:10000 myself,master - 0 0 1 connected 0-16383\n'
RESP = '${}\r\n{}\r\n'.format(len(CLUSTER_NODES_RESP), CLUSTER_NODES_RESP)

while True:
    connection, client_address = sock.accept()
    data = connection.recv(1000)
    if data == CLUSTER_NODES:
        connection.sendall(RESP.encode('utf-8'))
        raise Exception('Let it crash')

(2) Send a MGET SomeKey to corvus and trigger a crash.

[tevino]

LGTM

A test case would be great.