Reduce number of regions uptime-checks run in

[pnasrat]

See #2320

[pnasrat]

Tested on 2i2c staging via

terraform apply -target 'google_monitoring_uptime_check_config.hub_simple_uptime_check["staging.2i2c.cloud"]'

[consideRatio]

This LGTM!

As part of reviewing this, I tried catch up with my understanding about these uptime checks. Do you think this understanding is correct @pnasrat?

• We have "uptime checks" defined in the GCP project two-eye-two-see making HTTP(s) requests to check if something is up and running.
• Each uptime check has so far been a set of uptime checks, where the service checked is accessed from multiple locations of the world
• This change makes us only check service avaibility from USA
• This change is a change only to the two-eye-two-see GCP project, and not all individual cloud accounts with clusters.

[pnasrat]

@consideRatio see the docs here about the central project https://infrastructure.2i2c.org/en/latest/topic/monitoring-alerting/uptime-checks.html but yes. The main issue #2320 contains the detailed analysis of why (our free quota limit is being blown through) and this is an attempt to reduce costs until we can revisit monitoring and alerting design in general.

This is split off from the changing period due to the apparent need to destroy then apply to change period. I'm updating the docs when I change the period, and will add about region selection there when I get that out for review after testing if my assumption around the INVALID ARGUMENT error and period is correct.

Longer term I think these probes will be supplemented with detailed monitoring and alerting, that runs in cluster.