prompts: handle Runtime PM conditional gets

[OllieinCanada]

Fixes #396

Summary

Teach Sashiko's power-management prompt to account for the configuration-dependent return contract of pm_runtime_get_if_active() and pm_runtime_get_if_in_use().

Root cause

With CONFIG_PM=n, these helpers return a negative stub value while Runtime PM put operations are no-ops. Treating every negative return as an error that requires an early return can incorrectly suppress hardware operations in kernels built without Runtime PM.

Changes

• distinguish positive, zero, and negative conditional-get results
• require proof from the exact helper, configuration, and device lifecycle
• preserve reference-balance warnings when Runtime PM may be disabled for a device under CONFIG_PM=y
• avoid applying this exception to helpers with different contracts
• correct the existing IRQ example so it does not recommend ret <= 0 unconditionally

Validation

• verified the behavior against the upstream Linux Runtime PM header and implementation
• cargo fmt --all -- --check
• git diff --check
• exact one-file cumulative diff
• signed-off commit verified
• GitHub Actions DCO, lint, and Rust unit-test jobs pass

[rgushchin]

LGTM, but it's in a draft state, should I move on and merge it? Or there is something to wait for?

[OllieinCanada]

Thanks! Nothing else is pending from my side. I have marked the PR ready for review; please feel free to merge it.

[rgushchin]

Cool, thanks!
Can you, please, fix the SOB line with the real name and email?
Thanks

[OllieinCanada]

Updated the commit with my real author and Signed-off-by identity. DCO, lint, and unit tests are all green, and nothing else is pending from my side.