Migrate most tests to Jest

[lindapaiste]

Partially fixes #1330

Settings:

• Install jest and supporting packages (babel-jest, etc.)
• Install node polyfill for canvas.
• Add scripts to package.json to execute all .test.js files using Jest.
• Set env: { jest: true } in .eslint.js to fix es-lint errors re: undefined variable jest.
• Add testEnvironmentOptions: { resources: "usable" } to Jest config so that images will call the onload function.
  • This doesn't work for videos. Tests which load video still hang forever and are not included in this PR.
• Create a setupTests.js file in order to load the tfjs-node backend.
  • I tried to polyfill ImageData here, but it doesn't seem to work.

Utilities:

• Use utility functions asyncLoadImage and randomImageData instead of repeating code in multiple test files.
• Export a polyfillImageData function which can be used in a beforeAll function.

Test Modifications:

• Import the functions to test from local files instead of a global ml5 variable.
  • Rewrite test imports from const { charRNN } = ml5; to import charRNN from './index';.
  • import tf directly from @tensorflow/tfjs in NeuralNetwork instead of using ml5.tf alias.
• Replace all instances of jasmine.DEFAULT_TIMEOUT_INTERVAL = n; with jest.setTimeout(n);.
• Replace spyOn with jest.spyOn (which internally calls the jasmine spyOn function).
• Modify chained function calls, eg. .and.callFake to .mockImplementation, .and.returnValue to .mockReturnValue, etc.

Migrated Tests:

• BodyPix
• CVAE
• CartoonGAN
• CharRNN
• DBSCAN
• DCGAN
• FaceApi
• Facemesh
• FeatureExtractor
• Handpose
• ImageClassifier
• KMeans
• KNNClassifier
• NeuralNetwork
• ObjectDetector
• PitchDetection
• Pix2pix
• PoseNet
• Sentiment
• SketchRNN
• SoundClassifier
• StyleTransfer
• UNET
• UniversalSentenceEncoder
• Word2vec

The 57 skipped tests are all xdescribe placeholders in the NeuralNetwork file.

[joeyklee]

Wow! Thanks so much for this. I'd be totally happy to merge and release this so that we can add tests incrementally in case you think that would be helpful for you. Either way, this is great to see! 🏅