Implementation 805

[J007X]

This PR fixes [https://github.com//issues/805].

Description of changes

According to discussion and requirements in the ticket and , a new test NLP pipeline using Forte is created in a new test class "NLP_Pipeline_Performance_Test" to cover typical scenarios relating to entry/token, such as serialization, POS tagging and NER as mentioned in the ticket. Also to make sure 0.2.0 and current code base are run in exactly some way/code, some code in copied from some referenced (but changed over the version) packages, to make sure we perform an apple to apple test

Possible influences of this PR.

This can be used in several iterations of profiling-improvement efforts, that adding profiling coverage on key/typical scenarios.

Test Conducted

Profiling test on serialization, POS tagging and NER as requested in [https://github.com/J007X/forte/tree/implementation_805].

[codecov]

Codecov Report

Merging #905 (8068e13) into master (6e2d6ea) will decrease coverage by 0.07%.
The diff coverage is 72.48%.

@@            Coverage Diff             @@
##           master     #905      +/-   ##
==========================================
- Coverage   81.05%   80.99%   -0.07%     
==========================================
  Files         256      257       +1     
  Lines       19851    19999     +148     
==========================================
+ Hits        16091    16198     +107     
- Misses       3760     3801      +41     

Impacted Files	Coverage Δ
tests/forte/data/data_pack_profiling_test.py	72.29% <72.29%> (ø)
forte/data/ontology/ontology_code_generator.py	89.75% <100.00%> (ø)

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[hunterhector]

1. Maybe we also want to produce some numbers for the test?
2. How to run this test on actual data (like ontonotes), we shoud provide instructions.
3. The first line "PR fixes" message is not correctly formatted. Github won't associate the PR to the issue like this.