test: [M3-6632] - Add Machine Image region tests

[jdamore-linode]

Description 📝

This adds a handful of region tests (end-to-end) for Machine Image operations:

• Upload an Image via /images/create/upload
• Capture an Image from a Linode via /images/create/disk
• Update an Image's label and description from landing page
• Delete an Image from landing page

How to test 🧪

• Rely on automated Cypress tests to ensure that existing Image tests are not impacted
• Run the new tests with this command:

  CY_TEST_SUITE=region CY_TEST_REGION=us-east yarn cy:run -s "cypress/e2e/region/images/*.spec.ts"

(It takes roughly 5 minutes to run all of the tests against 1 region on my machine)

[mjac0bs]

Appreciated the background context from the volumes test sessions when looking at this. 👍🏼 Ran successfully and LGTM!

[mjac0bs]

Do you manually time these to get a rough estimate of how long these requests will take to complete?

[abailly-akamai]

@jdamore-linode there is a part of me that REALLY wants to mock this. These are pretty big timeouts

[jdamore-linode]

@mjac0bs Sort of! I'll usually run the tests a few times (just in the course of normal development) and get a rough gauge for how long these operations normally take, and in this case I set the timeout to be roughly double what I was observing.

@abailly-akamai Yeah, they are big :/ my hands are tied for the region tests, though -- there's not much value in using mocked data from a DC testing perspective, and machine image tests in particular are pretty important for the DC testing process.

I do want the timeouts to be high in this case because I want to be able to distinguish between cases where image uploading/processing is slow, and cases where it's broken completely. There are a couple mitigating factors that may hopefully make this less of a problem, though:

• The timeouts aren't really reflective of the length of time it takes for these operations to occur; from what I've seen, it usually takes 1-2 minutes to upload an Image, and 2-4 minutes to capture an Image from a Linode
• Right now the plan is for region tests to run only once daily or when manually kicked off (not against PRs*), so the run length of these tests shouldn't impact our day-to-day productivity
• As we develop slower end-to-end flows for the region testing pipeline, places where we have overlapping end-to-end test coverage in the core suite can be removed or replaced with mocked flows, hopefully giving us a lot of opportunity to improve the performance of our core suite

* All said and done, if the length of time it takes to run the region tests against a single region is roughly in line with our current test suite, then we could run region tests against PRs side-by-side with our existing test suite without it impacting our productivity, but it's not really on my radar right now

[abailly-akamai]

Looks good! Approving cause I don't have any alternative to adding timeouts other than mocking the response and I am not sure that's want you'd want to do if you want true e2e 🤷

Would be probably good to make a note of execution expectations it in the code for legacy?

[abailly-akamai]

@jdamore-linode there is a part of me that REALLY wants to mock this. These are pretty big timeouts