server settings for replication

[kickster97]

WHAT is this pull request doing?

Adds properties min_followers and max_lag to the Config class and the values are parsed from lavinmq.ini
The values are then conditioned in vhost.cr via the replication servers new wait_for_max_lag. You cannot publish to a queue if you have less followers connected than min_followers or if min_followers amount of followers have more lag than max_lag.

Fixes #674

HOW can this pull request be tested?

Start LavinMQ and configure the values in your config file, connect too few followers and try to publish.

[github-actions]

Main benchmark
'Average publish rate: 395624.7 msgs/s'
'Average consume rate: 395963.5 msgs/s'

PR benchmark
'Average publish rate: 403315.6 msgs/s'
'Average consume rate: 401376.7 msgs/s'

Keep in mind, these numbers are not representative of LavinMQ's peak performance.
It is rather an indication of how the changes of this pull request affects the performance of the main branch.

[kickster97]

@viktorerlingsson this is the solution i worked on with @carlhoerberg yesterday.

I added the wait_for_followers channel, which works if i take it slow with the followers, and don't have an agressive max_lag limit (around 10000b)
but gets into a bad state if the followers connect and disconnect too quickly when we have an agressive max_lag limit (like 100b)

[spuun]

👍

[kickster97]

Had to work around spec_helpsers after_each() that calls server.stop here.. because server.stop only calls replicator.clear, which will not finish if we are waiting for lag.