Fix ImportResourceState RPC response decoding for identity data

[austinvalle]

cc @dbanck

This PR fixes a bug in the most recent v1.12.0-alpha20250319 release (and used alongside the changes in #36703) when importing via resource identity.

Currently, the resource identity returned by ImportResourceState is being decoded with the resource's main schema instead of the identity schema, which will give you an error if they don't match, like:

examplecloud_thing.test: Preparing import... [identity=cty.ObjectVal(map[string]cty.Value{"id":cty.StringVal("id-123"), "name":cty.StringVal("austin")})]

Planning failed. Terraform encountered an error while generating this plan.

╷
│ Error: ["name"]: unsupported attribute
│ 

---

If you'd prefer to just fix this in #36703 (or if there is more to fix), feel free to just close this PR 👍🏻. I mostly just opened this so I could move forward 😆

---

Target Release

1.12.x

CHANGELOG entry

N/A

[dbanck]

Awesome! Thanks for catching this and your contribution.

I would like to have a test case for this, something like TestGRPCProvider_ImportResourceIdentity. Is this something you would be interested in contributing or should I add it to this PR?

[austinvalle]

Sure I can add that test here in a minute 👍🏻

[austinvalle]

Added a test to both protocol versions in 0d9a30e ⚡

I also adjusted the mock schema so that the identity schema is different than the resource schema + updated the upgrade test that relied on that schema 👍🏻

[austinvalle]

Hmm, the CI failed on a test that I can't reproduce on my local machine 🤔, seems unrelated to the changes I made here:

--- FAIL: TestUiHookPreApply_periodicTimer (3.01s)
    hook_ui_test.go:148: Output didn't match.
        Expected: "test_instance.foo: Modifying... [id=test]\ntest_instance.foo: Still modifying... [id=test, 00m01s elapsed]\ntest_instance.foo: Still modifying... [id=test, 00m02s elapsed]\ntest_instance.foo: Still modifying... [id=test, 00m03s elapsed]\n"
        Given: "test_instance.foo: Modifying... [id=test]\ntest_instance.foo: Still modifying... [id=test, 00m01s elapsed]\ntest_instance.foo: Still modifying... [id=test, 00m02s elapsed]\n"
FAIL

[jbardin]

time-based tests are notoriously flaky and need to be improved, re-running that will probably pass

[dbanck]

Thanks for adding the test!