Find better method for checking domain update

[rafaelgomesxyz]

We currently sleep for 30 seconds to wait for the domain to update when enabling/disabling maintenance mode. There should be a better way to do this to ensure we actually check if the domain was updated or not.

[justin808]

@rafaelgomesxyz will investigate.

[justin808]

@rafaelgomesxyz any update?

[justin808]

@zzaakiirr @rafaelgomesxyz #229

[justin808]

Triage assessment:

• Worth pursuing. The fixed delay is still present in MaintenanceMode#switch_domain_workload, where we call cp.set_domain_workload(...) and then Kernel.sleep(30).
• The repo now has retry/wait plumbing in Command::Base#step from PR Prevent infinite loop in Command::Base#step & add configurable wait time #217, so a focused fix can replace the blind sleep with a bounded poll that refetches the domain and checks that the route points at the target workload.
• This should include focused specs around the maintenance-mode switch behavior so we do not regress the existing maintenance:on / maintenance:off flow.

PR recommendation: open a small code PR for this.