Guidance for scheduled jobs needs updating

[TheRealJimShady]

What has been written to date is mostly still correct but the guidance could be expanded to include log and metric isolation as a key reason to favour jobs compute clusters over all-purpose compute clusters for scheduled jobs. Additionally, the instance pools feature is referenced as something that 'will' reduce startup time for short, frequent jobs - this has since been released into GA and should be reflected as such 🙂