PyTorch attention cells, transformer, bert and conversion script

[leezu]

Add PyTorch attention cells, transformer, bert and conversion script. Correctness verified via included tests and conversion script.

[github-actions]

The documentation website for preview: http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR1532/c52e64eedb4eb6623a4455f96ad62c6c0bdd783e/index.html

[github-actions]

The documentation website for preview: http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR1532/7459a3aa9eb30b50faf35df7eed6610b54a3ddd7/index.html

[github-actions]

The documentation website for preview: http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR1532/07aced5704fef8022408ea30bb615396320e9af8/index.html

[codecov]

Codecov Report

Merging #1532 (dbf8ce7) into master (223f1f6) will decrease coverage by 0.58%.
The diff coverage is 42.85%.

@@            Coverage Diff             @@
##           master    #1532      +/-   ##
==========================================
- Coverage   86.37%   85.79%   -0.59%     
==========================================
  Files          55       55              
  Lines        7522     7426      -96     
==========================================
- Hits         6497     6371     -126     
- Misses       1025     1055      +30     

Impacted Files	Coverage Δ
src/gluonnlp/utils/misc.py	54.09% <29.41%> (-2.26%)	⬇️
src/gluonnlp/__init__.py	100.00% <100.00%> (ø)
src/gluonnlp/base.py	100.00% <100.00%> (ø)
conftest.py	76.31% <0.00%> (-9.94%)	⬇️
src/gluonnlp/data/loading.py	76.55% <0.00%> (-2.28%)	⬇️
src/gluonnlp/utils/lazy_imports.py	58.42% <0.00%> (-2.25%)	⬇️
src/gluonnlp/data/tokenizers/yttm.py	81.73% <0.00%> (-1.02%)	⬇️
src/gluonnlp/data/tokenizers/spacy.py	65.33% <0.00%> (-0.91%)	⬇️
src/gluonnlp/data/tokenizers/huggingface.py	71.06% <0.00%> (-0.78%)	⬇️
... and 20 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 223f1f6...dbf8ce7. Read the comment docs.

[github-actions]

The documentation website for preview: http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR1532/dbf8ce79df53f68340a2de49b72d565a9c2a64ff/index.html

[sxjscience]

I'll approve. We can improve the testing later since this is the initial version of the pytorch-based GluonNLP.