[cDAC] GC Contract

[max-charlamb]

Implements and adds sub-descriptor as defined by #118126 for the GC. Each variation of the GC (SVR/WKS) has its own descriptor which is set up in PopulateDacVars.

Changes

• Add implementation mechanism for sub-descriptors in datadescriptor.cpp and cdac-build-tool
• Add new set of datadescriptors under src/coreclr/gc/datadescriptor mirroring the vm descriptors under src/coreclr/vm/datadescriptor.
• Additionally build the GC descriptors as part of coreclr.dll
• Bump gcinterface minor version to support adding new field gc_descriptor

Notes

In order to test the new GC descriptor, implements ISOSDacInterface::GetGCHeapData and ISOSDacInterface::GetGCHeapList.

• Due to constraints around defining explicit template specializations in a namespace that differs from the one that defined the template, I had to define the cdac_data specializations outside of the GC namespace (in datadescriptor.h). In theory we could move this into the GC namespace, but this conflicts with the existing definition used in the global namespace in coreclr. I thought reusing the same class and moving the specializations was reasonable.
• There are several times the GC is built without datadescriptors (or where they aren't supported). These include NativeAOT (something in bootstrap builds) and src/coreclr/gc/sample. We most likely need #ifdef out the extern specifiers in these scenarios.
  • Added GC_DESCRIPTOR define for builds when we expect the GCDescriptor.
• Currently waiting on updating coreclr to target c++17 Proposal: Upgrade C++ compiler feature level to C++ 17 #112419 to resolve build issues.

[risc-vv]

@dotnet/samsung Could you please take a look? These changes may be related to riscv64.

[dotnet-policy-service]

Tagging subscribers to this area: @steveisok, @dotnet/dotnet-diag
See info in area-owners.md if you want to be subscribed.

[noahfalk]

This looked fine to me but we should go over it with Maoni as well.

[mangod9]

GC side changes look good to me.

[noahfalk]

LGTM