SwiftArena and invoking destroy on objects as arena is destroyed

[ktoso]

Work in progress.

This will eventually allow us to bound lifetimes with a SwiftArena and sanitize that we don't "leak" objects in native by retaining them longer than intended.

[ktoso]

Moved some functionality out of SwiftKit, keeping it for things end users might wanna use.

[ktoso]

made these into tests rather than just one off prints in main()

[ktoso]

Heh I don't understand why the Java/Linux crash, local linux build is happy. I'll have to revisit this later

[ktoso]

Un-reproducible issues in CI which now manifest as hangs, quite a kerfuffle but I'll figure it out.

[ktoso]

So this reproduces only specifically in when running github actions, including using act so at least there's some hope to reproduce the issue. But it's pretty mysterious what's up with that.

[ktoso]

Specifically x86 has some issue with the witness table call and we just hang on:

   java.lang.Thread.State: RUNNABLE
        at java.lang.invoke.LambdaForm$MH/0x00007f25c31e6400.invoke(java.base@22.0.2/LambdaForm$MH)
        at java.lang.invoke.LambdaForm$MH/0x00007f25c31e8800.invokeExact_MT(java.base@22.0.2/LambdaForm$MH)
        at jdk.internal.foreign.abi.DowncallStub/0x00007f25c31e6800.invoke(java.base@22.0.2/Unknown Source)
        at java.lang.invoke.LambdaForm$DMH/0x00007f25c31e6c00.invokeStaticInit(java.base@22.0.2/LambdaForm$DMH)
        at java.lang.invoke.LambdaForm$MH/0x00007f25c31e7c00.invoke(java.base@22.0.2/LambdaForm$MH)
        at java.lang.invoke.LambdaForm$MH/0x00007f25c31e8000.invokeExact_MT(java.base@22.0.2/LambdaForm$MH)
        at org.swift.swiftkit.SwiftValueWitnessTable.destroy(SwiftValueWitnessTable.java:221)
        at org.swift.swiftkit.SwiftHeapObjectCleanup.run(SwiftMemoryResourceCleanup.java:37)
        at org.swift.swiftkit.ConfinedSwiftMemorySession$ConfinedResourceList.cleanup(SwiftArena.java:140)
        at org.swift.swiftkit.ConfinedSwiftMemorySession.close(SwiftArena.java:104)
        at org.swift.swiftkit.SwiftArenaTest.arena_releaseClassOnClose_class_ok(SwiftArenaTest.java:50)

trying to debug on a linux amd64 host, we'll see if I can find out what's up

[ktoso]

I disabled the hanging test on x86_64 so we can revisit it, I don't have good ideas what might cause that right now. We'll have to dig deeper into it specifically. It's logged as: #97

Unblocking other progress by merging this for now though.