Some update

[chunyuma]

This PR conducts the following:

1. Merge the recent update on the ARAX's algorithm from a closed PR that wasn't merged into the master branch.
2. Update the code/config_dbs.json to change the path location of new xDTD and xCRG database
3. Modify some tests for test_ranker.py

[dkoslicki]

Were the ranking results checked against the list from #2300? I.e. the comparison from here showed the new ranker was doing pretty well. Are the ranker changes in this PR already included in that? If not, it would be prudent to at least spot-check the ranker (better yet, run it through the benchmark repo) given these changes.

[chunyuma]

@dkoslicki, the change in this ranker script is not significantly different from the one in #2300, but just includes a looping algorithm proposed by Eric (see #2334). As I mentioned in Slack, it has passed almost all ranker tests but only failed on test_ARAXRanker_test9_asset619, test_ARAXRanker_test9_asset623, test_ARAXRanker_test13_asset355. I have listed the reason there. I think it is sufficient to demonstrate its validity, right？

[chunyuma]

BTW, reason of those three failed tests are not due to the change in the algorithm.

For test_ARAXRanker_test13_asset355, it seems to be due to the change in Node Synonymizer.

For test_ARAXRanker_test9_asset619 and test_ARAXRanker_test9_asset623, I am not sure if they are due to the latest version of xDTD. But based on what I have checked on the old results. the target drug results don't contain any inferred edge from xDTD so I guess they might come from the lookup function. However, the failed tests don't find the target results in the returned list. I am wondering if it is due to the KG2.10.0 problem.

[dkoslicki]

Yes, @chunyuma that does seem to be sufficient. I wanted to be sure this didn't negatively affect the good work you've done with the ranker. And the other issues seem non-ranker, non-xDTD, so I'll approve the PR