feat(iam): Add observe and update of user permissions boundary

[zonybob]

Description of your changes

This update causes the IAM User controller to also observe needed changes on the User permissions boundary and update it as needed. Part of this change also introduces the hopefully desirable side effect that UpdateUser is only called if needed (when the path changes.)

Fixes #1080

I have:

• [ X ] Read and followed Crossplane's contribution process.
• [ X ] Run make reviewable test to ensure this PR is ready for review.

How has this code been tested

This code has been tested against a factory environment where a number of IAM users are present and managed. Users were created, and boundaries were updated on the managed resource to observe the correct behavior in AWS.

IAM unit tests were also updated and run.

[zonybob]

I am working on updating this to support IAM Role as well (to fully close #1080 ). I'd appreciate any feedback/testing from the community as I do not use Roles in my setup. I will update unit tests and then test as much as I can.

[MisterMX]

Thank you for your contribution @zonybob.

This looks pretty close to merge for me. Can you fix the issues I remarked?

[zonybob]

@MisterMX thank you for the feedback!
I addressed error wrapping and UP_VERSION
I struggle a little bit with placement on IsUpToDate. I tried to follow typical pattern of placing in clients/iam, but I fight an import cycle with existing DiffIAMTags. Tried to stay out of refactoring others' work but can take your guidance.
Thanks!

[zonybob]

Also, @MisterMX you'll notice the failed e2e-tests
The only way I was able to get around that was upgrading up version to 0.16.1

[MisterMX]

notice the failed e2e-tests
The only way I was able to get around that was upgrading up version to 0.16.1

E2E Tests are running successfully for the latest master: https://github.com/crossplane-contrib/provider-aws/actions/runs/5508575184.

@zonybob you might have done some misconfiguration between build and the other make related files. Can you try running the branch without these changes?

[zonybob]

notice the failed e2e-tests
The only way I was able to get around that was upgrading up version to 0.16.1

E2E Tests are running successfully for the latest master: https://github.com/crossplane-contrib/provider-aws/actions/runs/5508575184.

@zonybob you might have done some misconfiguration between build and the other make related files. Can you try running the branch without these changes?

Yes I'm sorry! Forgot all about the build submodule. That is probably it. Will try that out.

[zonybob]

@MisterMX build submodule updated and e2e-tests passed. Let me know how it all looks now. Thanks!

[MisterMX]

Looking good for me except for some minor issues: Please encapsulate everything in a separate isUpToDate function and use a distinct error message for remote calls.

[zonybob]

@MisterMX thanks again for the response. New commits are up. I think I addressed everything this time (hopefully).

[MisterMX]

@zonybob that looks good now. THank you very much for your contribution!

One last thing: Can you please squash your changes into a single commit in order to keep the history short and concise? Then I can merge it.

[zonybob]

@zonybob that looks good now. THank you very much for your contribution!

One last thing: Can you please squash your changes into a single commit in order to keep the history short and concise? Then I can merge it.

@MisterMX are you able to squash on the merge? If not, could you assist me with this? My git log has intermingled commits from syncing master in periodically, and my rebase powers are minimal :)

[MisterMX]

@zonybob I could do half of that with Githubs "Squash and Merge" feature. However, it does not - for some reason - create a merge commit which is what we want as well. So we are left with squashing the commits manually.

Squashing in git is comparably easy and can be achieved with an interactive rebase. There are a bunch of tutorials out there that can help understanding it. Essentially it is just git rebase -i origin/master and replacing all feature commits pick (except the first) with either squash (s) or fixup (f).

[MisterMX]

Tested it locally and it seems only a single commit is merged anyways so it should be fine in this case.

[MisterMX]

Ok, it wasn't. Seems like Github is doing something differently under the hood when merging.