Add Mock test for IWSLT2017 dataset

[parmeet]

Reference issue: #1493

This PR do couple of things:

1. Add Mock test similar to IWSLT testing to start from compressed file #1596. Original credit: @erip due to Fix IWSLT2016 testing #1585
2. Fix an issue on IWSLT2017 that created additional texts folder outside of extracted archive

[parmeet]

@erip would be great if you can also take a look and catch any obvious issues :)

[codecov]

Codecov Report

Merging #1598 (ef3ab65) into main (18b61fa) will increase coverage by 2.02%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main    #1598      +/-   ##
==========================================
+ Coverage   82.11%   84.13%   +2.02%     
==========================================
  Files          58       58              
  Lines        2566     2566              
==========================================
+ Hits         2107     2159      +52     
+ Misses        459      407      -52     

Impacted Files	Coverage Δ
torchtext/datasets/iwslt2017.py	92.72% <100.00%> (+65.45%)	⬆️
torchtext/data/datasets_utils.py	64.74% <0.00%> (+5.75%)	⬆️

---

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 18b61fa...ef3ab65. Read the comment docs.

[Nayef211]

This PR looks good and seems consistent with the IWSLT2016 test. Since the two tests share so much similarity, do you think it would be feasible to parameterize them into one test class instead?

[parmeet]

This PR looks good and seems consistent with the IWSLT2016 test. Since the two tests share so much similarity, do you think it would be feasible to parameterize them into one test class instead?

hmm, not really. Their compression structures are quite different for us to efficiently merge the _get_mock_dataset function for the two test cases. The common part of them are already in datasets_utils.py.