Segfault on PHP5.5 and PostgreSQL

@alexpott: see #2871591-167: Allow ComplexData in TypedData to specify computed properties that should be exposed in normalization and other contexts, also @vaplas made it pass with second patch from #2879048-183: Ignore: patch testing issue for #2919863.
Also failing on SQLite (https://www.drupal.org/pift-ci-job/803921).

Hmm, the test runs seem to be failing all over the place with #2: https://dispatcher.drupalci.org/job/drupal_patches/36937/consoleFull, https://dispatcher.drupalci.org/job/drupal_patches/36936/consoleFull, https://dispatcher.drupalci.org/job/drupal_patches/36938/consoleFull, https://dispatcher.drupalci.org/job/drupal_patches/36939/consoleFull

Shoot. It looks like 5.5 test run has the same fail on #6: https://dispatcher.drupalci.org/job/drupal_patches/36959/consoleFull

18:00:38 Drupal\Tests\rest\Functional\EntityResource\EntityTest\Entit   0 passes             1 exceptions             
18:00:38 FATAL Drupal\Tests\rest\Functional\EntityResource\EntityTest\EntityTestJsonInternalPropertyNormalizerTest: test runner returned a non-zero error code (2).
18:00:38 Drupal\Tests\rest\Functional\EntityResource\EntityTest\Entit   0 passes   1 fails 

Re #9: the codebase isn't the same, your revert is missing #2871591: Allow ComplexData in TypedData to specify computed properties that should be exposed in normalization and other contexts.

@alexpott: my bad, I somehow assumed you ment to revert all of Nov 6.
The fail isn't 100%, e.g. for SQLite:
https://www.drupal.org/pift-ci-job/805468 is green
while
https://www.drupal.org/pift-ci-job/803921
https://www.drupal.org/pift-ci-job/804598
had the same problem

Looking at @vaplas patches it fails 2-8 times per 30 runs of affected tests, both SQLite and PostgeSQL.

Edit:
@alexpott: Retested #6 on PostgreSQL and it passed (https://www.drupal.org/pift-ci-job/806292).

@vaplas: #15 was meant to be a reply to #13.

#18 looks sensible.

Sure beats reverting important patches each time we step on unsupported PHP's toes, all just because of some vintage Ubuntu fetish ;).

The revert patch in the other issue is likely because it's affecting how the tests interact at high concurrency, however it's very likely to result in re-introducing previous segfaults we've already fixed. Moving this to RTBC.

@catch: that revert wasn't meant to be a solution, just a check to see if it will move the segfault out of EntityTest*JsonInternalPropertyNormalizerTest, it shows that there's nothing inherently wrong with #2871591: Allow ComplexData in TypedData to specify computed properties that should be exposed in normalization and other contexts itself.
+1.

#22: which revert patch? One of the patches in #2879048: Ignore: patch testing issue for #2919863? Reverting #2871591: Allow ComplexData in TypedData to specify computed properties that should be exposed in normalization and other contexts or #2871591: Allow ComplexData in TypedData to specify computed properties that should be exposed in normalization and other contexts would be a massive blow to the API-First Initiative.

Yep sorry that wasn't really for tacituseu just for anyone looking at this issue in 4 years who might get confused, but now everyone's confused now...

Committed/pushed to 8.5.x and cherry-picked to 8.4.x. Thanks!

🎉👏

This would also mean #2900399: Adding test coverage for batch bulk updates can go in now.