Revert default for GROWABLE_ARRAYBUFFERS to 0

[sbc100]

Split out from #27242.

See: #27241

[kripken]

Any idea why we missed these issues in our own testing? We check lots of WebGL and other Web APIs, did we just happen to not hit any bad ones?

[sbc100]

I guess because most of our tests don't use memory growth.

I think the test in #27242 should be enough to cover this.

[kripken]

I see, makes sense.

I wonder if we should add more growth testing? That mode is common enough that it possibly should be the default. If we have very little browser testing of it, that worries me (though maybe not enough for an entire new browser suite).

[sbc100]

I see, makes sense.

I wonder if we should add more growth testing? That mode is common enough that it possibly should be the default. If we have very little browser testing of it, that worries me (though maybe not enough for an entire new browser suite).

Yes, perhaps we should run the whole core tests with growable memory enabled? Or the entire browser tests suite?

[kripken]

Browser suite seems better for API coverage - but it also seems excessive. Not sure what's best here.