chore(shredder): propagate subplat deletes to cirrus

[relud]

Description

This PR adds shredder config for using fxa delete events to delete subplat cirrus pings,
and adds experiment monitoring config for experimenter_cirrus and subplat_cirrus

Related Tickets & Documents

• fixes CLONE - Step 8: Add subplat's deletion requests to cirrus shredder config experimenter#14106

Reviewer, please follow this checklist

[relud]

this is blocked on mozilla/probe-scraper#963 creating the targeted tables

[relud]

as I understand it, this is not blocked on https://github.com/mozilla-services/cloudops-infra/pull/6696, because the table has already been created

[mikewilli]

Typo fix but otherwise LGTM

[sean-rose]

I belatedly thought of a couple more things.

Also, I'm curious whether you've been able to test this ETL code yourself to verify it works end to end. If not, we'll probably want to be careful to deploy this at a time when you're available to quick-fix any issues that may arise.

[sean-rose]

question (blocking): Is it OK for this table to only have records from whenever it's deployed going forward?

If not, you'll probably want to output a different submission_timestamp condition for the table initialization query using Jinja and the is_init() function, including the appropriate historical range of dates.

[relud]

going forward is fine, we can truncate the impacted tables to avoid complicating this, esp knowing this is supposed to be "temporary" until some dev time is available after the EoY code freeze

[relud]

Also, I'm curious whether you've been able to test this ETL code yourself to verify it works end to end. If not, we'll probably want to be careful to deploy this at a time when you're available to quick-fix any issues that may arise.

I lack permissions to verify that getting secret values works as intended, I have manually confirmed the uid translation works as intended, as long as it's after 9am PST i'm available to fix

[sean-rose]

The deploy-changes-to-stage CI job failures are for materialized views, which often seem to be a problem when it comes to deployments. I'm not sure whether this is a known issue or some new breakage.

However, one easy way to unblock deployment of subscription_platform_backend_cirrus_derived.delete_events_v1 would be moving the experiment monitoring config changes to a separate PR.

[relud]

split out #8548