fix: Fixed signal issues

[abates]

This is an attempt to troubleshoot weird signal behavior in the CI pipeline that is not reliably reproduced locally.

[mzbroch]

I think we might be getting too complex here and we should improve modelling inside the bgp itself, shouldn't we ?

[abates]

I'm a little undecided about whether or not this functionality is too complex. There are clearly cases where we need to do something after an object has been saved. This is due to the fact that Nautobot commonly has business logic in the controller layer rather than the model layer. Specifically with BGP models we could move the peering updates to either the clean method or the save method in the model and it would solve this specific use case.

In the bigger picture, however, the question is: do we need to supply lifecycle hooks to action tags to handle these types of complex use cases?