Improve Schema For cc_set_passwords (SC-1105)

[holmanb]

Update chpasswd schema to deprecate password parsing

Additionally:

- chpasswd is specific to distro, make it a distro method
- add test to verify parity between new and old schema options
- update docs to encourage use of new schema keys

I've seen this come up as a concern multiple times from users. I'd also personally like to see this improved.

One remaining item to check before merging: what happens when identical users are in both list: and users:? Would it be better to catch this with a runtime exception or allow any error to bubble up from chpasswd? <- No longer a todo item: tests have been added that verify current behavior is preserved with the current change.

[TheRealFalcon]

Just a drive by comment...
I don't remember all of the details, but when #847 happened, we found a much larger use case for random passwords than we had expected. IIRC, it's even used by subiquity for installs. So we need to keep the random password feature. The new schema is nicer and involves much less guesswork, so +1 to that.

[holmanb]

So we need to keep the random password feature.

Sounds good. No problem, I'll add that to the new key as well.

[TheRealFalcon]

FYI, https://bugs.launchpad.net/cloud-init/+bug/1979065 . Not sure if it makes sense to address this here (I actually think it's probably better not to), but since you're in the middle of modifying some of this, you may want to incorporate this too.

[holmanb]

FYI, https://bugs.launchpad.net/cloud-init/+bug/1979065 . Not sure if it makes sense to address this here (I actually think it's probably better not to), but since you're in the middle of modifying some of this, you may want to incorporate this too.

I can do that in a follow up PR while this module is fresh in my brain.

[holmanb]

I just added some tests to verify that duplicate users in input don't break anything in a new way. This is arguably invalid input, but since we don't currently check for it I just wanted to ensure it didn't lead to brokenness while merging the two supported inputs together.

[holmanb]

The change here should be a noop change and was made for typing reasons.

[aciba90]

+1. Really nice UX schema addition.

I left you a minor nit. The indentation in that line was made with tabs instead of spaces.

[blackboxsw]

@holmanb much delayed review. but looks great. Just a missing deprecated: true I think and minor text changes.
Do what you will with the unittest suggestions.

Non-blocking comments, you can merge once deprecated: true is in schema.

[holmanb]

@blackboxsw - Great feedback, thanks! I committed your suggestions verbatim.

Non-blocking comments, you can merge once deprecated: true is in schema.

It looks like adding the deprecation key causes schemavalidation errors to be raised under strict validation in some tests.

You already gave approval, but given this change wasn't discussed previously I'll wait until you've had a chance to review that part (last commit).

[blackboxsw]

LGTM, and yes incorporates @aciba90's deprecation handling now from ccea0d4. - +1