metrics: add more metrics and fix some metrics

[djshow832]

What problem does this PR solve?

Issue Number: close #260

Problem Summary:
Currently, there are some problems of metrics:

• The avg duration and get backend duration is wrong
• We cannot inspect the latency from each tiproxy to each tidb, which may be affected by cross AZ network
• Lack of metrics that used for alert

What is changed and how it works:

• Fix avg duration and get backend duration
• Add ping duration from each tiproxy to each tidb
• Add server start event, load cert err, get backend fail counter, which are used for alert
• Other experience optimization, such as add more legends, change OPM to counter, etc.

Alert rules will be added in another PR.

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)
• No code

Notable changes

• Has configuration change
• Has HTTP API interfaces change
• Has tiproxyctl change
• Other user behavior changes

Release note

Please refer to Release Notes Language Style Guide to write a quality release note.

None