Make 'validate_onnx' function robust against the models loaded without external data

[dbogunowicz]

Accounting for the scenario, where the in-memory model passed to the validate_onnx is loaded without the external data.
Right now, the validator simply fails. What it should do, it should skip the validation and emit a verbose warning - this is what has been implemented.

Testing:

import onnx
from sparsezoo.utils.onnx import validate_onnx
model = onnx.load("/network/damian/sparsegpt_webinar_fp32/sparsegpt_1.3b/model.onnx", load_external_data=False)
validate_onnx(model)
model_ = onnx.load("/network/damian/sparsegpt_webinar_fp32/sparsegpt_1.3b/model.onnx", load_external_data=True)
validate_onnx(model_)

Attempting to validate an in-memory ONNX model that has been loaded without external data. This is currently not supported by the ONNX checker. The validation will be skipped.
Attempting to validate an in-memory ONNX model with size > 2000000000 bytes.`validate_onnx` skipped, as large ONNX models cannot be validated in-memory. To validate this model, save it to disk and call `validate_onnx` on the file path.

[rahul-tuli]

A few suggestions, but the diff looks real nice!

[KSGulin]

LGTM pending Rahul's comments