[release/8.0][browser] Fix failures in CalendarTestBase affecting several globalization tests

[ilonatommy]

Fixes #99378.
On net9 the fix was provided in #90881, this PR is a partial backport.

Description

The failures are caused by changing v8 version. HybridGlobalization relies directly on v8's functions output, so with the version change these tests are prone to break. Currently we test on Linux with v8=12.4.105.

Customer Impact

Users that called globalization APIs connected with time with HybridGlobalization on and ja-JP locale and are using v8 newer than 11.8 are affected.

Regression

Yes, with older v8 the output is correct.

Testing

Existing automated unit tests.

Risk

Small, it impacts only one specific use-case.

[carlossanlop]

I opened issues for the 3 the failures that showed up in the original CI run. They seem to be intermittent failures.

I also re-ran the failed legs and they now passed.

Just for my own peace of mind, @ilonatommy would you mind double checking this failure and confirm it's not related to this change or the one that caused this issue? #99423

[ilonatommy]

Just for my own peace of mind, @ilonatommy would you mind double checking this failure and confirm it's not related to this change or the one that caused this issue? #99423

Based on the issue history it seems to be OOM problem, e.g. #60234. I do not see how adding that "if-condition" block in the code could cause OOM + it's not the only occurrence of this failure. We can treat it as unrelated.

[carlossanlop]

Approved by Tactics via email.