Restricted Signature Aggregator L1s

[cam-schultz]

Context and scope
When an aggregate signature is requested from the signature aggregator from an L1 that it is not already tracking, it will first establish connections with that L1's peers before collecting signatures. The signature aggregator is able to be configured with an initial list of L1s to shift the peer connection step to startup rather than aggregation time.

We should provide a configuration option for the signature aggregator that prevents connections from being established with arbitrary L1 validator peers. This is important for controlled/compliant use cases where only some connections should be allowed. For this use case, the existing TrackedSubnetIDs configuration can be used to define the list of allowed L1s, and a new option restrictL1Connections would drop signature requests for L1s not in that list.

Open questions
How should we handle the primary network? One option would be to have a config option disallowPrimaryNetworkAggregation, but that may be overly specific.