circuit-relay-server limits and shades of transient

Nobody has claimed this yet.

- **Version**:
1.8.1

- **Subsystem**:
registrar
circuit relay server, registrar

#### Severity:
Medium or less

#### Description:
Since limits are applied to connection by default, all circuit relay connections will become transient. This means protocol handles that have set notifyOnTransient to `false` will not receive onConnect events for these connections.

But, are non-unlimited circuit relay connections really transient in the same way as the one we get from `identify`?
What is an intuitive way on listen to topology changes:

A. notifyOnTransient set to true, and filter out all transient connections except the relayed one
or
b. notifyOnTransient set to false, but set `applyDefaultLimit: false` on the circuit relay server config?

It feeeels, like `transient` could be replace with `limits` property object

and then every connection would have this property and we can use that to filter out unwanted connections from the protocol handler, for example one that will timeout to soon, or does not allow us to send that much data

Start by reading the circuit relay server and registrar handling described in the issue, then trace how connection limits, identify, and notifyOnTransient interact. Compare the applyDefaultLimit options and existing transient behavior before deciding whether the requested connection metadata and topology filtering have a defined scope; done requires an agreed design and corresponding behavior.

Written by the indexing model from the issue text.