nagios_uncommitted_crit default is very low, and alerts frequently

[Xav Paice]

In operations where the log volume is fairly high, we're seeing many alerts that resolve themselves shortly afterwards for the number of uncommitted messages. This is just that the application can't keep up with the load, but alerting for 100 messages is too low. We've had to increase this to 2000 or more in order to make the alerting meaningful.

[Tom Haddon (mthaddon)]

This has been included in cs:graylog-28