Refactor preprocess_cls to preprocess, add Serializer, add DataPipelineState

[ethanwharris]

What does this PR do?

Bit of a big one. Makes the following changes:

• Switch from {pre/post}process_cls to just handing around references to instantiated objects. This includes:
  • removing classmethods called e.g. instantiate_preprocess (now just handled in the __init__ of the class)
  • moving instantiation dependencies to the preprocess classes (e.g. default image transforms are now on the ImageClassification preprocess rather than the datamodule
• Adds a Serializer class so that a datamodule is now made up of a Preprocess, Postprocess, and a Serializer. These are classes which control the conversion from model output to desired prediction format (e.g. labels, classes etc.). They are seperate from the Postprocess as the user should be able to change the Serializer without messing with the more fundamental stuff in the Postprocess (without which the traning would usually fail). This Fixes Support more return types in predict #191 .
• Adds a DataPipelineState. Children of the data pipeline need to be able to communicate, this expands the PreprocessState to be shared across preprocess, postprocess, and serializer. This enables e.g. Labels serializer just gets the labels from the ClassificationState registered in the load_data of the ImageClassificationPreprocess.
• Adds an example showing multi_label image classification. Uses a subset of the movie poster genre data here: https://www.cs.ccu.edu.tw/~wtchu/projects/MoviePoster/
• Adds option to pass the data_fetcher as an argument to from_load_data_inputs

Before submitting

• Was this discussed/approved via a Github issue? (no need for typos and docs improvements)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure your PR does only one thing, instead of bundling different changes together?
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests? [not needed for typos/docs]
• Did you verify new and existing tests pass locally with your changes?
• If you made a notable change (that affects users), did you update the CHANGELOG?

PR review

• Is this pull request ready for review? (if not, please submit in draft mode)

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[codecov]

Codecov Report

Merging #229 (b1310e5) into master (b7436c4) will increase coverage by 0.24%.
The diff coverage is 92.13%.

@@            Coverage Diff             @@
##           master     #229      +/-   ##
==========================================
+ Coverage   86.57%   86.81%   +0.24%     
==========================================
  Files          58       58              
  Lines        2912     2981      +69     
==========================================
+ Hits         2521     2588      +67     
- Misses        391      393       +2     

Flag	Coverage Δ
unittests	86.81% <92.13%> (+0.24%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
flash/data/callback.py	98.70% <ø> (ø)
flash/data/data_module.py	77.92% <83.33%> (-0.16%)	⬇️
flash/vision/classification/data.py	88.20% <87.80%> (+0.34%)	⬆️
flash/tabular/classification/data/data.py	88.28% <88.46%> (-0.35%)	⬇️
flash/text/classification/data.py	82.47% <88.88%> (+0.65%)	⬆️
flash/core/classification.py	92.06% <90.90%> (-7.94%)	⬇️
flash/data/data_pipeline.py	87.87% <90.90%> (+0.59%)	⬆️
flash/core/model.py	91.42% <93.93%> (-0.06%)	⬇️
flash/data/process.py	88.73% <97.95%> (+1.99%)	⬆️
flash/data/batch.py	79.83% <100.00%> (+0.16%)	⬆️
... and 7 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b7436c4...b1310e5. Read the comment docs.

[kaushikb11]

@ethanwharris Could you please add some description about the PR.

[ethanwharris]

@kaushikb11 will do, only a draft at the moment 😃

[pep8speaks]

Hello @ethanwharris! Thanks for updating this PR.

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2021-04-22 10:02:35 UTC

[edgarriba]

neat tests :)

[tchaton]

Amazing PR :)