[MODEL][API] Change BaseTransformerEncoder to HybridBlock

[TaoLv]

Description

This PR intends to change the BaseTransformerEncoder to HybridBlock so we can export it without padding to fixed sequence length.
#789

TODOs:

• branch if arange_like is not available
• change export.py under scripts/bert/export/

Checklist

Essentials

• PR's title starts with a category (e.g. [BUGFIX], [MODEL], [TUTORIAL], [FEATURE], [DOC], etc)
• 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

❗ No coverage uploaded for pull request base (master@de73ce8). Click here to learn what that means.
The diff coverage is 88.88%.

@@            Coverage Diff            @@
##             master     #845   +/-   ##
=========================================
  Coverage          ?   79.65%           
=========================================
  Files             ?       66           
  Lines             ?     6379           
  Branches          ?        0           
=========================================
  Hits              ?     5081           
  Misses            ?     1298           
  Partials          ?        0

Impacted Files	Coverage Δ
src/gluonnlp/model/bert.py	64.84% <0%> (ø)
src/gluonnlp/model/transformer.py	81.36% <92.3%> (ø)

[codecov]

Codecov Report

❗ No coverage uploaded for pull request head (hybrid-block@702197e). Click here to learn what that means.
The diff coverage is n/a.

[eric-haibin-lin]

We can limit this PR to hybridize transformer encoder only. I can take care of the hybridization of BERTModel and the export script in a separate PR.

[TaoLv]

Thank you for working on this, @eric-haibin-lin . Sure, we can change the export script in another PR.

[eric-haibin-lin]

Fixed by #877