Finalize image classification preprocessing

[psyhtest]

We should update all image classification clients we support for MLPerf to use new preprocessing (in the order of priority - high to low):

• [x] ck-tensorflow:program:image-classification-tflite
• [x] ck-tensorflow:program:image-classification-tf-cpp
• [ ] ck-tensorflow:program:image-classification-tf-py
• [x] ck-mlperf:program:image-classification-onnx-py

Time-permitting, we should also update non-MLPerf clients:

• [x] ck-tensorflow:program:image-classification-tf-frozen-py
• [ ] ck-tensorflow:program:image-classification-tflite-py

At the same time, we should also:

• [ ] retire preprocess.py from ck-mlperf:script:image-classification.
• [ ] make the conversion script in ck-env:package:dataset-imagenet-preprocessed usable without CK.
• [ ] fully document ck-env:package:dataset-imagenet-preprocessed for use with and without CK.

[psyhtest]

Preprocessing should change the extension of files to e.g. rgb8 to avoid CK detecting original JPEG files as preprocessed ones.

Update: This required a corresponding change in postprocessing.

[psyhtest]

We should document the --ask flag:

$ ck install package --tags=dataset,imagenet,preprocessed --ask

(and perhaps enable it by default too).

[psyhtest]

For an AWS demo, we may want to use ck-tensorflow:program:image-classification-tf-frozen-py. We should raise its priority then.

[psyhtest]

As I understand, tf-cpp and tf-frozen-py have been updated?

[psyhtest]

With many changes to preprocessing over the last couple of months (and, importantly, finally matching the official code), this issue can be retired.