Add Engine.generate_random_inputs()

[mgoin]

An ease of use improvement to simplify demo code and reduce chance for errors.

Now we can do

from deepsparse import Engine

bert_engine = Engine(model="zoo:nlp/text_classification/obert-base/pytorch/huggingface/mnli/pruned90_quant-none")

inputs = bert_engine.generate_random_inputs()
output = bert_engine(inputs)

instead of

from deepsparse import Engine
from deepsparse.utils import generate_random_inputs, model_to_path

sparsezoo_stub = "zoo:nlp/text_classification/obert-base/pytorch/huggingface/mnli/pruned90_quant-none"
bert_engine = Engine(model=sparsezoo_stub)

inputs = generate_random_inputs(model_to_path(sparsezoo_stub), batch_size)
output = bert_engine(inputs)