policy-node: schedule pending rescan if old link is not active

This MR fixes a race condition that can happen in policy-node.lua. Currently, if a client node wants to be moved to a new device node before the first link completed, the policy removes the node while the old link is still being activated. This MR fixes this by scheduling another rescan if the previous link has not finished its activation, making sure only active links are removed.

This should fix "proxy activation failed" warnings such as:

wireplumber[126604]: <WpObjectManager:0x55976c7d9f20> proxy activation failed: No global specified; cannot bind proxy

Fixes #44 (closed)