Add configuration guide & generate script updates

[allisonlarson]

Adds an initial configuration guide with some basic use cases, highlighting the differences and any warnings about the config. This isn't meant to be exhaustive, but provides somewhere to start.

The Typical Values range in the chart is just what is in the guide, but it might not actually be super useful.

I wasn't sure how much information is useful to have here around priority calculation or usage & decay, but I can fill it in if its a good place for it.

The configurations have been tested manually, and its highlighted some possible additional configurations that could be helpful to allow for

• round-robin tenants by job count
• fairness based on resource & job count
• setting a lookback window that can be shorter than the gc value
• applying a penalty for a tenant submitting a bunch of jobs at once and missing the usage recalculations.

Also updates the generate jobs script to add randomization for cpu, mem, and priority (as well as some additional options to preserve existing functionality) for better testing of scenarios.

[allisonlarson]

I wasn't sure if these should be geared for testing (all durations should probably be short for observing behavior), or longer to show what a more realistic configuration might be?

[mismithhisler]

Yeah I think the calc_interval is just going to take tuning from users.. It probably also depends on how many jobs are queued up and how long those take to run.