gh-117482: Fix Builtin Types Slot Wrappers When Embedded

[ericsnowcurrently]

While gh-121602 solved most of the problem, it did not resolve the case where the runtime gets re-initialized. In that case, we end up in the same situation as before, where the inherited "tp" slots are set, which makes add_operators() do the wrong thing. We fix that by resetting the static type to the preserved values during finalization. We also make sure the type substructs are preserved.

(Given that this fix is for a more unlikely case, I'm inclined to not backport this to 3.12.)

• Issue: Weird int.__str__ behaviour inside sub-interpreters #117482

[AraHaan]

I feel like while we are at it, properly implementing a full cleanup of the runtime state as mentioned in the _PyRuntime_Finalize() function's comments and doing a full reinitialize of it's variables inside of _PyRuntime_Initialize(). I think doing that would help fix this issue permanently.

[ericsnowcurrently]

Yep. It's on my TODO list. 😄

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @ericsnowcurrently for commit 8520cfc 🤖

If you want to schedule another build, you need to add the 🔨 test-with-buildbots label again.

[freakboy3742]

!buildbot iOS

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @freakboy3742 for commit 8520cfc 🤖

The command will test the builders whose names match following regular expression: iOS

The builders matched are:

• iOS ARM64 Simulator PR

[ericsnowcurrently]

I'm going to go with gh-122865. We can circle back to this later.