Pass cmd.run arguments to onlyif/unless cmd execution

[mchugh19]

What does this PR do?

Extends the global onlyif/unless cmd handling to accept arguments defined in the cmd.run state.

What issues does this PR fix or reference?

Fixes: #57760

Previous Behavior

Previous to #55974, the cmd.run state handled its own unless and onlyif support. This included passing any of the cmd.run arguments for the state into the onlyif and unless cmd as well (https://github.com/saltstack/salt/pull/55974/files#diff-6617770c26f25c18a6adeeaf4ccdebd3L336)

When this behavior was removed in favor of using the global onlyif/unless requisites this behavior was not included, so arguments like shell: /bin/bash which were defined for cmd.run were not honored by the cmd.retcode execution used by onlyif/unless.

New Behavior

Arguments accepted by cmd.run (which make sense) are now passed to cmd.retcode used by onlyif and unless.

• cwd
• root
• runas
• env
• prepend_path
• umask
• timeout
• success_retcodes

Other possible arguments such as name, output_loglevel, or creates, don't make sense for the unless/onlyif cmd.retcode use, while the arguments bg and use_vt would break functionality.

Support of shell is a special case, and will be pulled first from the state arguments, then from the value of the shell grain if either exist.

Risks: it is possible that generic terms like cwd, env, or root, may exist in other states with different meanings than is intended for the cmd.retcode use in onlyif/unless. If desired, a gate could be added to optionally ignore inclusion of arguments to cmd.retcode, but has not yet been included as I don't know if the risk is worth the additional option complexity.

Merge requirements satisfied?

[NOTICE] Bug fixes or features added to Salt require tests.

• Docs
• Changelog - https://docs.saltstack.com/en/master/topics/development/changelog.html
• Tests written/updated

Commits signed with GPG?

No

[jdelic]

From reading this patch it brings back env for onlyif. Removing the setting of environment variables for cmd.run requisites has broken my whole setup across multiple machines and required a lot of manual fixes.

Can this land as a (in my case urgent) bugfix for 3001 in the short term?

[krionbsd]

@mchugh19 LGTM, could you please run pre-commit run and re-submit the patch?

[Silvenga]

I would also agree with @jdelic - this breaks a lot for us. PowerShell is a lot different from cmd.

Can we get this backported to 3001?

[mchugh19]

@krionbsd @garethgreenaway since pre-commit ran the py2 cleanup process, and this touches state.py, it'd be nice to get lots of eyes on this.