improve on handling job data from a different master_job_cache

[erwindon]

When a different master-job-cache is used, e.g. postgres, then some details are different:

• runner/wheel jobs are now also visible in the jobs list. they have a stored target field which holds an artificial value (hostname + _master).
• when viewing the jobs page, for runner/wheel jobs show an error because it does not know how to handle the return codes from these. it is visible as 1 failure(undefined)
• when viewing the jobs page, for the regular jobs no minions is show. this is what started this github issue. for runner/wheel jobs that is actually correct, but maybe also superfluous.
• some of the internally used runner/wheel commands are not hidden from the jobs lists. the policy in SaltGUI is to hide internally used commands so that the more interesting jobs are visible.

SaltGUI should improve on handling those situations by compensating/explaining/ignoring.
This issue is the result of the work for #427

[erwindon]

• all job types used by SaltGUI are now hidden (again). added runner.manage.versions, runner.doc.wheel, runner.doc.runner and wheel.minions.connected.

[erwindon]

• jobs without return codes now show a neutral summary

[erwindon]

• not solving the "no minions" effect as that is normal for runner/wheel jobs

[erwindon]

• not solving the strange "target" value

we have no good ways to predict these values as for (hostname + _master), we need to know the hostname of the master. that is information we do not have.
also this is not really bad

[erwindon]

@pckroon fyi