[Metric] Add new metrics to monitor TCP status

[morningman]

Motivation

When we operate and maintain the cluster ourselves, we have encountered many times that the state of the Doris cluster is abnormal due to machine network failures, such as heartbeat failure and query blocking.

Because there is no corresponding metrics, it takes a long time to troubleshoot the problem.

New metrics

Add 2 new metrics

tcp_in_errs
tcp_retrans_segs

These 2 metrics is got from /proc/net/snmp, indicate the number of error tcp request and number of retransmit tcp requests.

[chaoyli]

If you have add TCP error, you can also add more error as follow.

atmptf/s
estres/s
retrans/s
isegerr/s
orsts/s

You can get the manual from https://linux.die.net/man/1/sar.
And also, you can use the following command to get the result.

sar -n ETCP 1

[morningman]

If you have add TCP error, you can also add more error as follow.

atmptf/s
estres/s
retrans/s
isegerr/s
orsts/s

You can get the manual from https://linux.die.net/man/1/sar.
And also, you can use the following command to get the result.

sar -n ETCP 1

In our recent operation and maintenance experience, only these two parameters are currently instructive. Other parameters, such as atmptf/s, are always in a non-zero state, which is not helpful for troubleshooting. If there is a demand later, we can add it later.