Improve performance of extract_related()

[juyrjola]

Check more thoroughly if the related resource is already included in the response, and skip serialization if it is. For my workload with a lot of included resources, this optimization decreased the list endpoint processing time by around 50 %, because extract_related() was repeating serialization for many of the resources that were already included in the response.

[codecov]

Codecov Report

Merging #532 into master will decrease coverage by 0.52%.
The diff coverage is 66.66%.

@@            Coverage Diff             @@
##           master     #532      +/-   ##
==========================================
- Coverage   96.11%   95.58%   -0.53%     
==========================================
  Files          56       55       -1     
  Lines        2859     2856       -3     
==========================================
- Hits         2748     2730      -18     
- Misses        111      126      +15

Impacted Files	Coverage Δ
rest_framework_json_api/renderers.py	86.94% <66.66%> (+0.14%)	⬆️
rest_framework_json_api/metadata.py	25% <0%> (-23.34%)	⬇️
example/tests/test_filters.py	100% <0%> (ø)	⬆️
example/settings/dev.py	100% <0%> (ø)	⬆️
example/tests/integration/test_pagination.py	100% <0%> (ø)	⬆️
example/tests/integration/test_polymorphism.py	100% <0%> (ø)	⬆️
example/models.py	100% <0%> (ø)	⬆️
example/tests/test_serializers.py	100% <0%> (ø)	⬆️
rest_framework_json_api/django_filters/backends.py	100% <0%> (ø)	⬆️
example/tests/unit/test_utils.py	100% <0%> (ø)	⬆️
... and 8 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 5650bd1...dd2506c. Read the comment docs.

[sliverc]

Thanks for your contribution. Could you add a test that covers both continue statements (also see codecov)?

This will show that this improvement works.

[sliverc]

@juyrjola Are you still interested in working on this PR? A test is missing as a last task.

[sliverc]

As there hasn't been any response to #532 (comment) concerning a test and proving this improvement I am closing this for now.

Please feel free to reopen if you continue working on it though.