Update dashboards to support Helm chart deployment

[danfromtitan]

What this PR does:

Update cortex-mixin to support Helm chart deployment and produce usable dashboards. List of issues fixed by this change:

• support disabling cortex-gw, query scheduler and ruler panels
• generalize instance names to also match Helm chart deployments
• remove cluster from namespace matcher to cope with metrics outside Cortex service that don't have a cluster label
• fix disk panels to display correct written byte metrics
• support custom Prometheus http prefix
• display read/write metrics from Cortex distributor/querier instead of cortex-gw
• rearrange panels in rollout dashboard to make better sense of the metrics shown

Caveat:

• I tested dashboards with Helm chart deployments but since that uses opensource Cortex image I could not test the dashboards when cortex-gw component is enabled.

Which issue(s) this PR fixes:
Fixes #361

Checklist

• CHANGELOG.md updated - the order of entries should be [CHANGE], [FEATURE], [ENHANCEMENT], [BUGFIX]

[jmcarp]

Do you all have capacity to review this @pracucci? Would be useful for my team as well.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

---

Dan Constantinescu seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}