Fix incorrect OtherContextException wrapper when the exception comes back to its creator context

[eregon]

Summary

Fixes #13453

Testing

I wrote a test for this in truffle/src/com.oracle.truffle.api.instrumentation.test/src/com/oracle/truffle/api/instrumentation/test/TruffleContextTest.java with the help of Codex. It passes with the fix and fails without.

I also have tests in TruffleRuby for this: https://github.com/truffleruby/truffleruby/pull/4235/changes
I confirm this PR fixes #13453, as in those tests fail without and pass with this PR.

Documentation

No documentation updates are needed.

Contributor Checklist

• I have read the contribution guide.
• I have the right to contribute the submitted material under the project terms.
• I have updated tests and documentation where appropriate.
• If I used a coding assistant, I remain responsible for the entire contribution and have reviewed it accordingly.

[eregon]

migrateValue() does:

graal/truffle/src/com.oracle.truffle.polyglot/src/com/oracle/truffle/polyglot/PolyglotContextImpl.java

Line 1731 in 3c0e0b0

return new OtherContextGuestObject(this, otherValue.delegate, valueContext);

(in migrateValue(), this == receiverContext)

Which is correct?
I'm not sure what the receiverContext is for, wouldn't only tracking the delegateContext be enough?
Is it because the OtherContextGuestObject/OtherContextException objects need to belong to one context and not many? (if so, is that needed?)

[eregon]

I (and Codex) have figured this out, it was another bug, this time in migrateValue().
Pushed a commit with a fix + test and verified the test passes with the fix and fails without.

[oubidar-Abderrahim]

Thank you for the PR @eregon
@chumer could you please review, thanks

[eregon]

@chumer @tzezula I wrote a test now with the help of Codex (and reviewed it), and also addressed the comment above which turned out to be a second bug, so now this PR fixes 2 bugs and adds a test for each. Please review.

[eregon]

Ah during rebasing I noticed @iamstolis already fixed the second bug in 5b5c93a / #13522.
I'll drop my last commit then as that's already fixed & tested.

[tzezula]

Looks good to me.
It unwraps the object originated in the targetContext rather than creating a decorator chain.

[chumer]

lgtm. inbounding the PR. sorry for the delay.

[chumer]

PR is merged here: #13599
(will reuse your branch next time, to track this properly, sorry)