docs: clarify terminology, scope, and architecture model in README

[mindpower]

Summary

This PR updates the README to resolve ambiguity regarding the definitions of "Agent" vs. "Server" and explicitly defines the distinct scopes of the AICard and AICatalog specifications. It introduces the term "AI Service" as the standard logical entity and renames the protocol map to "Interfaces" to avoid naming collisions.

Motivation

Recent discussions (e.g., #6, #4 (comment), #4 (comment)) highlighted that:

1. The term "Agent" is too narrow for MCP tool servers, while "Server" is too ambiguous (confusing Host vs. Entity).
2. Using services inside the Card to list protocols created a naming collision if we ever wanted to refer to the top-level entity as a "Service."

Key Changes

1. New Terminology Hierarchy:
   • Host: The infrastructure/domain.
   • AI Service: The logical entity (covers both autonomous Agents and passive Tools).
   • Protocol Interface: The specific endpoint (A2A/MCP) exposed by the Service.
2. Scope Clarification:
   • AICard: Defined as the Identity & Capability definition (The "Digital Passport").
   • AICatalog: Defined as the Host Inventory (The "Building Directory").
3. Schema Alignment:
   • Renamed the protocol map in the Card description from services to interfaces to align with A2A's supportedInterfaces and prevent naming conflicts.
4. Discovery Patterns:
   • Clarified support for both Decentralized (.well-known) and Centralized (Registry API) discovery.

Impact

These changes align the documentation with the consensus reached in the schema discussions in V1 schema proposal PR, providing a clearer mental model for implementers integrating both A2A and MCP protocols.