Replicas continue to acquire more files for reading even when they are already processing a file, while some other replicas aren't.

Example:

Reading 100 files on a cluster with 100 replicas having this work distribution:

https://pastila.nl/?0000ebe6/7efa107d36a0080e5d387dabd83cfa90#vtB0IbU26F9OyU9jj0W2xQ==

This issue is a copy of https://github.com/ClickHouse/clickhouse-private/issues/10615