Simplify methods

[characat0]

Right now most of the methods expect to receive instances of MLFlow, MLFlowExperiment and/or MLFlowRunInfo.
This makes calls unnecessarily verbose, as opposed to the python client, that manages a global state.

From our docs

using MLFlowClient

# Create MLFlow instance
mlf = MLFlow("http://localhost:5000/api")

# Initiate new experiment
experiment_id = createexperiment(mlf; name="price-paths")
exprun = createrun(mlf, experiment_id)

logmetric(mlf, exprun, "pricepath", pricepath)

From the MLflow Tracking QuickStart Guide

import mlflow

# Set our tracking server uri for logging
mlflow.set_tracking_uri(uri="http://127.0.0.1:8080")

# Create a new MLflow Experiment
mlflow.set_experiment("MLflow Quickstart")

# Start an MLflow run
with mlflow.start_run():
    # Log the hyperparameters
    mlflow.log_params(params)

    # Log the loss metric
    mlflow.log_metric("accuracy", accuracy)

The Python API is more concise, and the context manager also takes care of finishing the run even if the code inside fails.
If its something desirable, I can start working on adding extra methods for simplifying our API.

[pebeto]

Hello @characat0.

The approach of MLFlowClient.jl is just to be a client, enabling direct communication to the mlflow instance REST API. This approach is followed to enable other ML platforms, just like MLJ.jl, to use it and take care of the context (see MLJFlow.jl).
Also we are trying to avoid "black magic" by stay pure functional.

I'm open to discuss this if you have a concern.

[characat0]

It is mostly syntactic sugar tbh, the same can be accomplished by a wrapper, maybe MLJFlow.jl is more suited for this discussion as you suggested.