Replace cython sampler with mxnet sampler to improve user experience

[eric-haibin-lin]

Description

Now we have _sample_unique_zipfian in mxnet, we don't need to build with cython.
Ran the training script end2end for 33 epochs, reached test ppl 44.15.

Checklist

Essentials

• Changes are complete (i.e. I finished coding on this PR)
• All changes have test coverage
• Code is well-documented

Changes

• Feature1, tests, (and when applicable, API doc)
• Feature2, tests, (and when applicable, API doc)

Comments

• If this change is a backward incompatible change, why must this change be made.
• Interesting edge cases to note here

[codecov]

Codecov Report

Merging #270 into master will decrease coverage by 5.32%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #270      +/-   ##
==========================================
- Coverage   75.12%   69.79%   -5.33%     
==========================================
  Files          74       75       +1     
  Lines        6202     6277      +75     
  Branches      964      988      +24     
==========================================
- Hits         4659     4381     -278     
- Misses       1319     1697     +378     
+ Partials      224      199      -25

Impacted Files	Coverage Δ
scripts/tests/test_scripts.py	96% <100%> (-4%)	⬇️
scripts/nmt/train_transformer.py	0% <0%> (-73.95%)	⬇️
gluonnlp/data/dataloader.py	26.08% <0%> (-69.57%)	⬇️
scripts/nmt/loss.py	42.37% <0%> (-18.65%)	⬇️
scripts/nmt/transformer.py	84.67% <0%> (-6.52%)	⬇️
scripts/nmt/translation.py	82.05% <0%> (-5.13%)	⬇️
scripts/nmt/utils.py	75% <0%> (-3.58%)	⬇️
scripts/tests/test_sampler.py	100% <0%> (ø)
...cripts/language_model/large_word_language_model.py	0.41% <0%> (+0.41%)	⬆️
scripts/language_model/sampler.py	75.55% <0%> (+75.55%)	⬆️
... and 1 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 07b0edd...e4ca873. Read the comment docs.

[mli]

Job PR-270/8 is complete.
Docs are uploaded to http://gluon-nlp-staging.s3-accelerate.dualstack.amazonaws.com/PR-270/8/index.html