[EPIC] SDK Architecture Modernization & Thread-Safety Refactor

[fderuiter]

Objective

Resolve foundational architectural flaws in the SDK to ensure production readiness, thread-safety, and maintainability. This epic tackles the entanglement of async/sync paradigms, eliminates critical concurrency hazards in caching, modernizes the plugin system, and aligns the monorepo with modern Python tooling standards.

Scope of Work

This epic tracks the following distinct implementation phases:

• Phase 1 (Critical): # [Insert Issue ID] - Remove stateful caching from endpoint classes to resolve thread-safety vulnerabilities.
• Phase 2 (Architecture): # [Insert Issue ID] - Decouple Async and Sync client hierarchies.
• Phase 3 (Extensibility): # [Insert Issue ID] - Standardize plugin architecture for imednet-workflows using importlib.metadata.
• Phase 4 (Tooling): # [Insert Issue ID] - Migrate monorepo configuration from custom workspace.toml to native Poetry 2.0 workspaces.
• Phase 5 (Tech Debt): # [Insert Issue ID] - Remove implicit/dynamic kwargs parsing in GenericListGetEndpoint.

Definition of Done

• All linked sub-issues are closed.
• Core test suite passes with 100% coverage on refactored modules.
• Documentation is updated to reflect the new strict separation of Async/Sync clients.