[hyper-v] Invalidate cached mgmt IP address on update_state()

[xmkg]

The cached management IP address becomes stale on some operations when it happens out-of-band, like "multipass restart", where it's a SSH-induced guest reboot.

This patch fixes that by invalidating the cached IP address on every update_state() invocation.

Fixes: #3354

MULTI-2068

[ricab]

Many thanks for (re-)spotting and fixing this @xmkg!

The general idea makes sense to me, although in the long run I hope we can find a way to be alerted to ACPI events in the guest. I suspect it would be possible to subscribe over a VSOCK/SSH connection, otherwise we might use a heartbeat based approach.

I have a few requests inline, but I think we should let this one simmer in main for a while either way. It's been like this forever and the fix could have effects that we're not anticipating, so I wouldn't rush to release it now.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 89.31%. Comparing base (b7a35a3) to head (5088e5b).
Report is 2 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #4207   +/-   ##
=======================================
  Coverage   89.31%   89.31%           
=======================================
  Files         259      259           
  Lines       15684    15684           
=======================================
  Hits        14008    14008           
  Misses       1676     1676           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.