Add drawIo diagrams

Signed-off-by: JeffMboya <jangina.mboya@gmail.com>

Author: JeffMboya

docs/fml.md

@@ -38,6 +38,8 @@ Federated learning provides natural scalability advantages:
 
 Propeller's FML system is built on a workload-agnostic design where the core orchestration layer (Manager) has no FL-specific logic. Instead, FL-specific functionality is handled by an external Coordinator service that manages rounds, aggregation, and model versioning. This separation of concerns allows Propeller to support federated learning while remaining flexible enough to orchestrate other types of distributed workloads.
 
+![The FML system consists of the following components](images/fml-architecture.png)
+
 ### Core Design Principles
 
 1. **Workload-Agnostic Manager**: The Manager service orchestrates task distribution and message forwarding without understanding FL semantics. It treats FL tasks like any other workload, creating tasks and forwarding messages verbatim.