BF: lock across threads check/instantiation of Flyweight instances

[yarikoptic]

As commented added to the comment says we might want (eventually) make locking smarter and per "id" since in principle we should be able to instantiate multiple separate instances in parallel. Might be useful or even critical for parallel "create" and alike. But I think such use-cases are rare so decided to KISS. test added, so disable locking to see how things could go wrong (hint: there is more than 1 way ;) ).

Fixes #6598

[qlty-cloud-legacy]

Code Climate has analyzed commit 8eaffc4 and detected 0 issues on this pull request.

View more on Code Climate.

[codecov]

Codecov Report

Base: 74.92% // Head: 74.92% // No change to project coverage 👍

Coverage data is based on head (dc90cc9) compared to base (dc90cc9).
Patch has no changes to coverable lines.

❗ Current head dc90cc9 differs from pull request most recent head 4fb3bfd. Consider uploading reports for the commit 4fb3bfd to get more accurate results

Additional details and impacted files

@@           Coverage Diff           @@
##            maint    #7075   +/-   ##
=======================================
  Coverage   74.92%   74.92%           
=======================================
  Files         354      354           
  Lines       58945    58945           
  Branches     6310     6310           
=======================================
  Hits        44164    44164           
  Misses      14767    14767           
  Partials       14       14           

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[bpoldrack]

Thx!

While I am not entirely sure that this was the issue #6598, it seems possible and this locking should be an improvement either way.

[yarikoptic]

While I am not entirely sure that this was the issue #6598, it seems possible and this locking should be an improvement either way.

I would leave 10% case that indeed it was something else since I do not remember if that test should have somehow instantiated the same dataset at different time points. But from the signature of the failure, and the fact that we get exactly that error of introduced test if you remove locking -- I am almost certain it is the same issue, so I will close it there.

[yarikoptic-gitmate]

PR released in 0.17.7