Increase network read timeout in GitLab Migration

According to Kibana logs, Net::ReadTimeout frequently occurs during migrations in GitLab Migration. Although we have a retry mechanism when such errors happen, it seems sensible to increase the read timeout and give more time for the source instance to reply.

Tech detail

We should increase the network read timeout only for the requests via Sidekiq. For synchronous requests, we need to keep it as is to avoid the request taking longer than the rack timeout.