Update errors field in JSON Report

[mehrinkiani]

This PR updates errors and skipped field to have a category, description and source (optional for error) field for JSON reporting as outlined in #107.

Close #107

Sample result for a h5 model that does not have a model config:

    "summary": {
        "total_issues_by_severity": {
            "LOW": 0,
            "MEDIUM": 0,
            "HIGH": 0,
            "CRITICAL": 0
        },
        "total_issues": 0,
        "input_path": "tf_model.h5",
        "absolute_path": "/Users/mehrinkiani/Documents/modelscan",
        "modelscan_version": "0.0.0",
        "timestamp": "2024-02-22T15:59:25.994786",
        "scanned": {
            "total_scanned": 0,
            "scanned_files": []
        },
        "skipped": {
            "total_skipped": 1,
            "skipped_files": [
                {
                    "category": "MODEL_CONFIG",
                    "description": "Model Config not found",
                    "source": "tf_model.h5"
                }
            ]
        }
    },
    "issues": [],
    "errors": []
}
}

Sample results for a dependency error:

    "summary": {
        "total_issues_by_severity": {
            "LOW": 0,
            "MEDIUM": 0,
            "HIGH": 0,
            "CRITICAL": 0
        },
        "total_issues": 0,
        "input_path": "notebooks/KerasModels/unsafe_model.h5",
        "absolute_path": "/Users/mehrinkiani/Documents/modelscan/notebooks/KerasModels",
        "modelscan_version": "0.0.0",
        "timestamp": "2024-02-23T12:17:35.614094",
        "scanned": {
            "total_scanned": 0
        },
        "skipped": {
            "total_skipped": 0,
            "skipped_files": []
        }
    },
    "errors": [
        {
            "category": "DEPENDENCY",
            "description": "To use modelscan.scanners.H5LambdaDetectScan, please install modelscan with h5py extras. `pip install 'modelscan[ h5py ]'` if you are using pip."
        }
    ]
}

[swashko]

Thinking this should be a skipped reason instead of an error. In the benchmarking we're encountering .bin files that aren't pytorch and we should just be skipping them instead of erroring

[mehrinkiani]

Thank you @swashko for sharing this! I have now updated it to add them to skipped files

[willarmiros]

just some small style things, in general lgtm!

[willarmiros]

LGTM