The mnist model is used in multiple tests for minimal/mobile builds in both ONNX and ORT formats.

We also save the ORT format version of the model with level 1 (aka 'basic') optimizations applied: mnist.basic.ort.
This file is used in NNAPI and CoreML EP unit tests.

It can be generated with the following steps:
- Set environment variable ORT_CONVERT_ONNX_MODELS_TO_ORT_OPTIMIZATION_LEVEL=basic
- From this directory, run
  $ python -m onnxruntime.tools.convert_onnx_models_to_ort --optimization_style=Fixed ./mnist.onnx
