[Post 0.6][Tabular] Speed up feature transform in tabular NN model

[liangfu]

Description of changes:

This PR simplifies the feature transform function in tabular NN model.

1. avoid using string-based processing
2. leverage numpy.isin(b, a, invert=True) function instead of running np.array([item in b for item in a])

Benchmark results are shown below

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Job PR-2442-25f53bd is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2442/25f53bd/index.html

[liangfu]

cc @tonyhoo @Innixma

[Innixma]

Looks awesome, I see an even bigger speed up of around 40% using m6i.16xlarge in my testing!