Optimize the performance of slow_query and log-search In many files scenarios

[crazycs520]

Development Task

When the tidb_slow_query.log and tidb.log files more than 1000, query CLUSTER_SLOW_QUERY and CLUSTER_LOG will be very slow. This issue uses to optimize the performance in this scenario.

Bugfix:

• Fix cancel log-search as soon as possible. *: update go.mod for sysutil #21373
• Fix cancel query SLOW_QUERY as soon as possible. *: fix cancel query slow_query doesn't release resource and tiny optimize #21456.

Performance optimization:

• Change read slow-log file module to concurrent Change read slow-log file module to concurrent #20890
• optimize query cluster_slow_query order by time desc when there are many slow-log files optimize query cluster_slow_query when there are many slow-log files. #20236
• Add log file meta cache. @crazycs520 executor: add slow-log file meta cache to avoid repeat read file meta information #21536

[zimulala]

Most of the tasks have been completed, except for the last one. However, the last one hasn't been dealt with yet. Let's close this issue for now.
If the last optimization point still needs to be addressed, we can create a separate issue later.