Proxy and Reflect implementation

[rbri]

Next try on the long story.... see #268

This is the follow up of #1431.

[p-bakker]

Not possible to continue with the previous PR? This way we may lose the (unresolved) comments

[rbri]

As promised i have recreated the Proxy/Reflect stuff based on the current status of the impl.
There are still a bunch of failing test but i can't move forward without breaking backward compatibility - already discussed at other places.

From my point of view this is already in a state that can be merged - i will not really work on this because i have some other priorities. Will try to rebase from time to time....

Status

	head	Proxy & Reflect
built-ins/Array	383/3074 (12.46%)	365/3074 (11.87%)
built-ins/ArrayBuffer	142/191 (74.35%)	140/191 (73.3%)
built-ins/BigInt	21/75 (28.0%)	20/75 (26.67%)
built-ins/Boolean	4/51 (7.84%)	3/51 (5.88%)
built-ins/DataView	254/550 (46.18%)	250/550 (45.45%)
built-ins/Date	90/770 (11.69%)	87/770 (11.3%)
built-ins/Error	6/41 (14.63%)	5/41 (12.2%)
built-ins/Function	186/508 (36.61%)	184/508 (36.22%)
built-ins/JSON	37/144 (25.69%)	26/144 (18.06%)
built-ins/Map	13/171 (7.6%)	12/171 (7.02%)
built-ins/Math	51/327 (15.6%)	16/327 (4.89%)
built-ins/NativeErrors	31/123 (25.2%)	23/123 (18.7%)
built-ins/Number	24/335 (7.16%)	23/335 (6.87%)
built-ins/Object	222/3408 (6.51%)	189/3408 (5.55%)
built-ins/Promise	406/631 (64.34%)	392/631 (62.12%)
built-ins/Proxy		79/311 (25.4%)
built-ins/Reflect		13/153 (8.5%)
built-ins/RegExp	1169/1854 (63.05%)	1168/1854 (63.0%)
built-ins/Set	167/381 (43.83%)	166/381 (43.57%)
built-ins/String	140/1182 (11.84%)	127/1182 (10.74%)
built-ins/Symbol	36/94 (38.3%)	35/94 (37.23%)
built-ins/TypedArray	1091/1422 (76.72%)	1084/1422 (76.23%)
built-ins/TypedArrayConstructors	597/735 (81.22%)	566/735 (77.01%)
built-ins/WeakMap	15/102 (14.71%)	14/102 (13.73%)
built-ins/WeakSet	12/85 (14.12%)	11/85 (12.94%)
language/expressions/typeof	2/16 (12.5%)	0/16 (0.0%)
language/statements/for-of	453/741 (61.13%)	452/741 (61.0%)

[p-bakker]

There are still a bunch of failing test but i can't move forward without breaking backward compatibility - already discussed at other places.

Could you elaborate on which breaking changes are needed to make Proxy more spec-compliant? From looking at all comment on this and the two preceding PRs, I gathered the following:

Included breaking change: defineProperty now returns boolean instead of void
Excluded breaking change: changing Scriptable.delete to return boolean

Are there more changes that ought to be made, but that you held off on as they would constitute a breaking change?

Also, some of my previous comments remain unanswered. Any chance you can have a look at them?

• implementation of Reflect & Proxy support #1431 (comment) (just the part about invariant checks)
• implementation of Reflect & Proxy support #1431 (comment)
• implementation of Reflect & Proxy support #1431 (comment)

And wrt to whether this is mergeable in its current state or not: I worry that there are details of the spec that haven't been implemented yet and that do not relate to requiring breaking changes, for example (some completely random checks on currently failing tests):

• https://github.com/tc39/test262/tree/main/test/built-ins/Reflect/get/return-value-from-receiver.js: unless there's something fundamentally wrong in the login of looking up a property value up the prototype chain in Rhino already, I would think this would be an implementation detail of Reflect that isn't quite correct yet
• https://github.com/tc39/test262/tree/main/test/built-ins/Reflect/set/receiver-is-not-object.js: Seems a Reflect impl. detail, unless this is an auto-boxing-related issue in Java?
• https://github.com/tc39/test262/tree/main/test/built-ins/Reflect/set/return-false-if-receiver-is-not-writable.js & https://github.com/tc39/test262/tree/main/test/built-ins/Reflect/set/return-false-if-target-is-not-writable.js: Also seem Reflect impl details, unless there are fundamental flaws in PropertyDescriptors on plain JS Objects in Rhino
• https://github.com/tc39/test262/tree/main/test/built-ins/Proxy/set/trap-is-null-receiver.js also looks like an implementation detail

Not trying to be pedantic about this, I really wanna see Proxy support merged (heck, I opened the issue about it 8 and a half years ago :-) ), am just worried about A: merging something that isn't as spec compliant as it can be, meaning we'll have to make (potentially) breaking changes later and B: that if we don't take care of the details now, we'll never get to tme