Improve HA tracker documentation

[pracucci]

Today I supported an user on Slack to correctly setup the HA tracker and I captured a lot of good feedback to improve the HA tracker documentation (Slack thread):

• HA tracker documentation is sparse between arguments page and its dedicated page
• $POD_NAME mentioned in the arguments page is vague (env variable substitution is not supported by Prometheus)
• We should suggest how to deploy two Prometheus servers in HA pairs with different external labels
• Having two Prometheus servers with different external labels generates duplicated alerts in the alertmanager and we should explain how to fix it (see the Slack thread)

[pracucci]

As mentioned in #3572, we should also document on which metrics to look out to try to guess if HA tracker is correctly setup.