Change Pytorch materializer to support remote storage

[safoinme]

Describe changes

change PyTorch materializer input & output handler to support remote storage using zenml.io fileio context manager.
this will allow saving model artifacts directly to any storage system.

        with fileio.open(
            os.path.join(self.artifact.uri, DEFAULT_FILENAME), "wb"
        ) as f:
            torch.save(model, f)

I have added another save for the model checkpoint which is essential for the inference phase.

Pre-requisites

Please ensure you have done the following:

• I have read the CONTRIBUTING.md document.
• If my change requires a change to docs, I have updated the documentation accordingly.
• If I have added an integration, I have updated the integrations table.
• I have added tests to cover my changes.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Other (add details above)

[htahir1]

If you have tested it works on remote backends then I'm happy to approve. Would be cool if you could test on GCP and AWS at least and Azure would be a bonus :-)

[safoinme]

If you have tested it works on remote backends then I'm happy to approve. Would be cool if you could test on GCP and AWS at least and Azure would be a bonus :-)

@htahir1 I have added an other save in pytorch materializer to save entire_model and checkpoint. using the entire model cause error in inference (something related to torch.load)
for testing, I have done it for GCP and it's working. I'm almost done with Pytorch custom deployment with Seldon once it's finished I will ask @stefannica to test it in AWS since it's his main stack I think.

[bcdurak]

Great changes. I left a tiny comment, but other than that everything looks good to me!

[bcdurak]

If we don't need to do anything here, could we perhaps remove the last two lines?

[safoinme]

yes sure, it's removed!