Feature/SK-1102 | Simplify ClientAPI and add support for inference

[niklastheman]

Added a readme file to the client directory. This is an attempt to make users see how easy they can create their own clients.

Benjamin:
Key changes

• Moved boiler plate code from client to ClientAPI to make it simpler for the user to implement an on_train/on_validate function
• Updated client_v2 accordingly
• These changes eliminate the client's (client_v2 or exemple in README) need to import fedn protos
• Added support for predict in ClientAPI and GrpcHandler
• Added functions in GrpcHandler to create gRPC messages for model update, validation and prediction

grpc_handler.py

• Added function create_update_message, create_validation_message and create_prediction_message which populate and return a gRPC message of the relevant type
• Function send_model_update() now takes a ModelUpdate message as input directly and only handles logic for sending the message to the server

client_api.py

• Replaced _subscribers with train_callback and validate_callback
  • train_callback needs to take in_model as parameter and return out_model and metadata dict
  • validate_callback needs to take in_model as parameter and return metrics dict
• _task_stream_callback now invokes update_local_model or validate_global_model depending on task request type
  • update_local_model and validate_global_model contain boiler plate code that was previously inside the on_train and on_validate callbacks (getting/sending model from/to combiner, generating uuid, assembling metadata, sending model update/validation)
  • This replaces the functions train and validate
• run method creates a child thread for heart beats and listens to task stream in main thread
• gRPC messages for update, validation and predict are created in update_local_model, validate_global_model and predict_global_model respectively and are sent in status and to GrpcHandler

client_v2.py

• Adjusted training and validation functions for the updated client_api.py

README.rst

• Updated example client that makes use of the new API

[ahellander]

Nice, I think it will be helpful to have something like this.

However I think we can make it even neater, I have some ideas, let's talk about it.

[niklastheman]

Nice, I think it will be helpful to have something like this.

However I think we can make it even neater, I have some ideas, let's talk about it.

Yes, let's!

[ahellander]

I think it is better to give a minimal, but working example. This could be confusing.

[ahellander]

Don't understand this. How can the client set the round_id?

[ahellander]

Remove printouts?

[benjaminastrand]

• Added predict to GrpcHandler and ClientAPI
• Added functions to create gRPC messages for update, validation and prediction
  • ClientAPI sends request + model data to GrpcHandler and receives gRPC message back
  • gRPC message is sent to combiner and in status
• Updated README (minimal example of ML training will be added in a new issue)

[Wrede]

Imo I think you should change the Name of ClientAPI to maybe Project, ProjectManager, Context, ContextManager and import it into the root init so that you can:
from fedn import Context

[benjaminastrand]

Changed name of client api

• ClientAPI -> FednClient
• client_api.py -> fedn_client.py