Add setting s3_slow_all_threads_after_network_error

Why it matters

This feature addresses the issue of multiple threads aggressively retrying a network error simultaneously when reading from S3, which can overwhelm the service or network. By slowing down all threads after a network error, it helps to reduce retry bursts and improve overall stability and reliability during transient network failures.

How to use it

Enable the feature by setting the s3_slow_all_threads_after_network_error configuration option to true in your ClickHouse server settings. This will cause all threads reading from S3 to slow down after encountering a network error, mitigating stress on the network and S3 service during retries.