BAAS-20006: Update goja to latest (March 2023) - Part 2

[Gabri3l]

Conflicts

Implemented classes (including private elements).

• Added call.ctx to builtin_function.go on line 198. It won't look like a change because I ported it over from before.
• Updated signature of f._call to now take a context param too. This was necessary because we don't pass call as an argument anymore but we pass the function call arguments directly.
• Updated common_eval signature base on the new eval function signature.
• Generally moved MemUsage functions around, nothing particularly crazy.

Fixed "delete 'a'.prop". Optimised a[<const_string>] to a.<const_string>

The “array with non-numeric keys" needed a change. There was an optimization introduced with dot notation, before we had two string props:

• x["a"] = 3
• x["c"] = 3

Both were the last item added on the stack before we check the memory the only difference between the two mem checks was the added "abc" and the added number.

After the change this is not true anymore and elements that can use dot notation are not added to the stack anymore. This simply means that "abc" is now the last item on the stack before we call the mem check again which would have resulted in additional mem increase of 3 (i.e. len("abc")). To "even things out" the last item added before the mem check are assigned with a number key value which will isolate again "abc" and the added number so the resulting mem increase stays the same.

Fixed Export() value for arrow functions and classes. Added Runtime.A…

• There was some refactoring due to some shared code between AssertFunction and AssertConstructor which was encapsulated in runWrapped. I just had to make sure we were still passing the correct context along.

Refactored conversion to primitive. Always use String() method if d…

• Had to remove a few unused methods in object_goreflect and object_goslice

BAAS

We have an evergreen patch here that's looking promising. The failing tests seem unrelated.

[kpatel71716]

LGTM! Some qs but no blockers

[LijieZhang1998]

I'll blindly approve it cause I don't want to block it.

[jwongmongodb]

Thanks for breaking up to the commits. I don't have much to add for the PR, just a few optional cosmetic fixes since it came from the original goja code

[jwongmongodb]

[opt] we can group the no-ops together in the switch statement

[jwongmongodb]

[opt] can combined this with the above check

if this && derivedCtor{
...
}

[Gabri3l]

Both very legitimate changes! I'm only NOT going to do it to minimize conflicts in the future. Although our diversions are not clear yet (working on it ...) those are the only places where we can 100% make these type of changes. Much appreciate either way!