BGP: Add operator policy to define maximum prefix count and violation action

A customer recently encountered a situation where their upstream BGP peers (controlled by a different administrative entity) increased the number of prefixes they were advertising to their Oxide system, which surpassed the number of routes supported by the Switch ASIC.

This brings about the question of how best to handle this situation (which is likely best addressed from multiple angles).
One approach would be to allow an operator to define upper limits on the amount of prefixes it is willing to receive (per AF).

One could imagine a scenario where this could be configured per-peer or globally, as well as offering different actions to be taken when the prefix limit is surpassed (e.g. resetting the BGP session, discarding routes received beyond the upper limit, alerting from the routing daemon, etc.).