Clarify management and cpu interface type descriptions

[dplore]

Change Scope

• Add more detail to the descriptions for cpu and management interface types.
• Since the component tree for management and cpu interfaces both should trace up to a CPU component, more description and examples are helpful to ensure the intended types are used.

[OpenConfigBot]

No major YANG version changes in commit 31b8562

[LimeHat]

I think connected to the CPU phrasing is imprecise. They're co-located on the same board, but the CPU (a chip) is not a direct owner of the interface and does not have a direct connection. If we start counting indirect connections, then any regular interface is connected to the CPU.

[dplore]

Perhaps all we can say that is unique about a management connected interface is:

1. it is not connected to an INTEGRATED_CIRCUIT.
2. It is externally accessible to the device for user management access.

Is this more clear and precise enough to disambiguate from a 'cpu' interface?

[LimeHat]

That seems reasonable.
FWIW I'm happy with the current (upstream) definition of the management leaf. In my view, ambiguity was related mostly to the definition of the cpu leaf (and the way it is used in gnmi-1.22 https://github.com/openconfig/featureprofiles/blob/4cff2c8fa1815887a481b0b051fbf077464c47ad/feature/platform/controllercard/tests/port/README.md?plain=1#L32-L33)