Use saved plan to determine CreateBeforeDestroy status

[jbardin]

If the planed action changes from CreateThenDelete to DeleteThenCreate,
terraform will present an error like:

Error: Provider produced inconsistent final plan

When expanding the plan for null_resource.a to include new values learned so
far during apply, provider "registry.terraform.io/hashicorp/null" changed the
planned action from CreateThenDelete to DeleteThenCreate.

This is a bug in the provider, which should be reported in the provider's own
issue tracker.

This was being reported to providers as the message indicated, so wasn't
immediately visible to core developers, e.g.

hashicorp/terraform-provider-aws#15044
hashicorp/terraform-provider-aws#15048
https://discuss.hashicorp.com/t/this-is-a-bug-in-the-provider-which-should-be-reported-in-the-providers-own-issue-tracker/13190
hashicorp/terraform-provider-aws#13823 (comment)
hashicorp/terraform-provider-aws#10297 (comment)

Since the provider does not influence the status of CreateBeforeDestroy,
we first fix the output to indicate that this is a bug in terraform.

In order to prevent the error, we need to check for a saved plan with a
CreateThenDelete action and rely on that to determine the status of
CreateBeforeDestroy. Due to the eval-node pattern currently in use,
this needs to be maintained in both the apply node and the eval nodes.
Future refactoring can work on unifying these codepaths, and making
these internal flags less fragile.

Fixes #26180

[codecov]

Codecov Report

Merging #26192 into master will decrease coverage by 0.01%.
The diff coverage is 40.00%.

Impacted Files	Coverage Δ
terraform/eval_diff.go	68.22% <35.71%> (-1.25%)	⬇️
terraform/node_resource_apply_instance.go	80.31% <100.00%> (+0.15%)	⬆️
internal/providercache/dir.go	70.68% <0.00%> (-5.18%)	⬇️
dag/marshal.go	53.33% <0.00%> (-1.12%)	⬇️

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.