`pm-suspend` shouldn't be used to trigger suspend

### Enhancement Proposal

I was recently browsing the tests and came across https://github.com/canonical/checkbox/blame/main/providers/base/bin/sleep_test.py#L147 using `pm-suspend` for triggering suspend cycles.

`pm-utils` is long dead upstream and modern Ubuntu installations use systemd to trigger suspend cycles (ie `systemctl suspend`). The routine that systemd uses is different than `pm-suspend` in that systemd will trigger the lock screen and also cause displays to use DPMS to turn off.

The sequence of events that pm-utils uses doesn't guarantee this, and it could lead to difficult to identify sequencing bugs in GPU drivers. As Ubuntu natively uses systemd, it would be better to migrate the test suite to actually test using systemd.

Inspect providers/base/bin/sleep_test.py around line 147 and read the surrounding suspend-cycle test. Check how the test suite invokes the current suspend command, then run the relevant sleep test; done means the cycle is triggered through Ubuntu's systemd path rather than pm-suspend.

Written by the indexing model from the issue text.