feat: add monthly_hrs usage field for aws/azure/gcp VMs

[jessecureton]

Updates the following resource types to support a monthly_hrs usage flag:

• aws_instance
• azurerm_virtual_machine
• azurerm_linux_virtual_machine
• azurerm_windows_virtual_machine
• google_compute_instance

This addresses #1556.

[CLAassistant]

All committers have signed the CLA.

[jessecureton]

At its core this is a pretty straightforward PR, but the myriad golden file updates have made it appear larger than it is.

Happy to break this up into a separate PR each for AWS / Azure / GCP resources, if the maintainers prefer. Looking forward to getting feedback and getting this landed!

[aliscott]

Thanks for the contribution @jessecureton 🙏! One of our team will look at this in the next couple of days.

[jessecureton]

Sounds good! I took a look at the failing test case in the CI checks and it seems unrelated to this change. I did notice the failing test case (TestBreakdownWithPrivateTerraformRegistryModule) does not run locally, and instead prints the following message.

breakdown_test.go:369: Skipping because INFRACOST_TERRAFORM_CLOUD_TOKEN is not set and external contributors won't have this.

Locally running make test_all passes all tests as expected, so hazarding a guess this seems like it might be something related to CI secrets when running in my forked repo?

[aliscott]

@jessecureton you're correct. I pushed a fix to skip this test in #1809, but I think there might still be issues here I haven't had a chance to investigate today.

[aliscott]

Thank you @jessecureton, this looks great 🙂. Two minor points below.

1. I think we should we use monthly_hrs to be consistent with the other *_hrs usage keys.
2. Should it be a float value for the use-case if a user only runs the instance for a few minutes in the month?

[jessecureton]

1. I think we should we use `monthly_hrs` to be consistent with the other `*_hrs` usage keys.

I used monthly_hours to match the same field name already in use in several other resources - namely azurerm_synapse_spark_pool, aws_glue_job, and aws_glue_crawler. That said, I'm happy to make this change if you still prefer the _hrs syntax, just wanted to point out the mixed convention already in use.

2. Should it be a float value for the use-case if a user only runs the instance for a few minutes in the month?

This seems reasonable! I will make this change alongside the above once I hear back re: the preferred syntax

[aliscott]

I used monthly_hours to match the same field name already in use in several other resources - namely azurerm_synapse_spark_pool, aws_glue_job, and aws_glue_crawler. That said, I'm happy to make this change if you still prefer the _hrs syntax, just wanted to point out the mixed convention already in use.

Yeah we definitely have inconsistencies at the moment 😞 and at some point we should migrate those to be consistent. For now we should use _hrs to be consistent with the resource mapping guidelines.

[jessecureton]

I've updated this to include both the requested changes for float values and the switch to monthly_hrs! Thanks for the link to the policy doc for naming, I had missed that initially!

[aliscott]

Thanks @jessecureton, this looks great! I'll rebase and run the tests, to make sure they all pass.