move to non-deprecated sql exporter

[mrjones-plip]

Following the recent work done in medic/cht-watchdog#81, this PR

• updates compose file to stop using the SQL exporter from deprecated one (prometheuscommunity/postgres-exporter) and start using the supported one (burningalchemist/sql_exporter)
• converts queries from old YAML format to new
• updates scrape files in prometheus
• updates JSON dashboard files in grafana

per #112

[mrjones-plip]

test steps

1. in watchdog repo, check out branch mrjones-sql-exporter-data-ingest-dont-merge
2. in watchdog, make sure you have a symlink data-ingest -> ../cht-app-monitoring-data-ingest/watchdog-config
3. in data ingest, checkout branchmrjones-migrate-sql-exporter
4. in data ingest, copy watchdog-config/sql_servers_example.yml to watchdog-config/sql_servers.yml
5. in data ingest edit sql_servers.yml to have a valid postgres username and password and IP based off your local RDBMS tunnel
6. in data ingest, edit scrape.yml so that scrape_interval: 10s and scrape_timeout: 5s - don't commit these values though!
7. in watchdog ./development/kill.start.ips.sh

demo video of test steps

data-ingest-exporter-demo.webm

demo mapping new metrics to old metrics

taking the dwh_replication_by_status metric, we can open the panel in the "Edit" view and compare it to the live metric on Watchdog. We can get metric parity by:

1. selecting the the new metric: replication_by_status_replication_failure_count -> dwh_replication_by_status
2. adding another label filter: type = replication_failure_count

old: last_over_time(replication_by_status_replication_failure_count{cht_instance="$cht_instance"}[$__interval])
new: last_over_time(dwh_replication_by_status{cht_instance="$cht_instance", type="replication_failure_count"}[$__interval])