metrics: record tidb_server_handle_query_duration_seconds inaccurate

[srstack]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

1. set status.record-db-label=true
2. use test
3. select * from mysql.user;
4. query metric tidb_server_handle_query_duration_seconds from prometheus

2. What did you expect to see? (Required)

tidb_executor_statement_total{db="mysql", instance="127.0.0.1:10080", job="tidb", type="Select"}

3. What did you see instead (Required)

tidb_executor_statement_total{db="test", instance="127.0.0.1:10080", job="tidb", type="Select"}

4. What is your TiDB version? (Required)

nightly