[mlflow flavor] Distributed of MLForecast doesn't support for load model

[Aeim]

What happened + What you expected to happen

When I use mlforecast.flavor to load distributed version of model, it show the error that No such file or directory: '/local_disk0/repl_tmp_data/ReplId-19203-35577-8/tmpta101dhm/model/mlforecast-model/ts.pkl' because the model of distributed is save in name as _base_ts.pkl It may be change it to run to be success.

Versions / Dependencies

mlforecast==0.13.4

Reproduction script

loaded_model = mlforecast.flavor.load_model(model_uri=model_uri)

Issue Severity

Medium: It is a significant difficulty but I can work around it.

[jmoralez]

Hey. Whats your expectation here? For the distributed model to be saved into a local one? Or support loading it for distributed inference?