consolidate tag and tags diff during asg update

[enxebre]

Because of #3532:
After importing and applying an asg, Terraform tries to destroy and recreate tags with same values yielding into additional aws permissions needed to manipulate tags, which are not needed for creating the asg.
This consolidates "tag" and "tags" at the time of making the diff to build the payload for the aws request.
Note that this is intended to be an improvement to avoid an unneeded extra aws request, but the root problem in #3532 is still terraform import generated stated always setting tag field even if the config contains tags

[bflad]

Hi @enxebre 👋 Thank you for contributing this and sorry it never got a proper review in the meantime. The Auto Scaling service tags are certainly a pain. 😖

We plan on fixing the aws_autoscaling_group resource tags in two ways:

• Centralizing the conversion and difference logic to match all other AWS services that support resource tagging to remove the custom code, see also: provider: Support arbitrary additional tag data, implement keyvaluetags and ignore_tags support in aws_autoscaling_group resource, return empty strings with key-only tags instead of panicking #13868
• Deprecating the tags set of maps argument (an argument type rarely, if ever, used anywhere else in the Terraform AWS Provider and was only present in this resource as a workaround for Terraform 0.11 and earlier configurations) in preference for the tag configuration blocks. This will allow us to consolidate all the logic to a single difference operation.

The combination of these should supersede the intent of this pull request, as #13868 should cover a similar difference collapse and the future deprecation will allow the aws_autoscaling_group resource tagging to behave like all other resources without needing to handle two tagging arguments. Importing and state saving of the tags will only be in the single schema attribute.

Please reach out if you have any questions and thank you again for submitting this.

[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 feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. Thanks!