Different jobs require different resources. Sometimes these collide.

What I would like to see introduced is jobs to have an input field where group
names are written.

So when job "X" is in the queue, it is NOT executed until there are no more jobs
executed in one or more collision groups of job "X".

Example:

queue - job "X" in collision groups "mysqlperf" and "net1perf"
currently executed - job "Y" in collision groups "net1perf" and "ftpperf"

So job "X" must wait for job "Y" to complete before being executed.
---------
I see an issue for the matrix project where it would be cool to have the ability
to specify collision groups per axis. It is highly likely for jobs on the same
axis to collide. So the user should specify the collision groups of every axis.