[cDAC] Implement DacDbi heap walking

[Copilot]

• Cleans up the native DAC implementation of heap walking. Upon E_FAIL, a consumer of this API keeps trying and trying again to advance the heap walk, hoping that we will encounter a good object or switch to the next segment. Refactored the implementation such that we switch to the next good segment upon reaching heap corruption instead of striding forward by the last known object size - hopefully improving performance and discovering more objects on corrupt heaps. We do need the intermediate E_FAIL to signal to consumers "this is a partial heap".
• Implements the heap walking algorithm in cDAC, including a linear cache for performance in the large heap enum loop, matching the native implementation .
• Implements some GC-heap related alignment helpers:

    // Returns the next candidate object address within a heap segment.
    TargetPointer GetPotentialNextObjectAddress(
        TargetPointer currentAddress,
        ulong currentObjectSize,
        GCHeapSegmentInfo segment) => throw new NotImplementedException();

    // Aligns an object's raw size to the alignment required by its containing segment.
    ulong AlignObjectSize(ulong size, GCSegmentClassification generation) => throw new NotImplementedException();

• Implement APIs CreateHeapWalk, WalkHeap, and DeleteHeapWalk

[dotnet-policy-service]

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

[Copilot]

Pull request overview

This PR extends the cDAC-based DacDbi implementation to support GC heap walking, adds GC contract helpers needed to mirror the native DAC walking logic (object-size alignment and allocation-context skipping), and refactors the native DAC heap walker to advance to the next valid segment/object on corruption while still signaling partial-heap results via E_FAIL.

Changes:

• Implement CreateHeapWalk, WalkHeap, and DeleteHeapWalk in the managed (cDAC) DacDbiImpl, including a linear read cache and per-segment enumeration.
• Add IGC.GetPotentialNextObjectAddress and IGC.AlignObjectSize (and implement them in GC_1) to support heap-walk stepping/alignment parity with native.
• Refactor native DacHeapWalker to advance to the next valid object/segment after corruption and fix allocation-context bookkeeping.

Show a summary per file

File	Description
src/native/managed/cdac/Microsoft.Diagnostics.DataContractReader.Legacy/Dbi/DacDbiImpl.cs	Adds managed heap-walk handle lifecycle + iterator-based walking with a small page cache and alloc-context handling.
src/native/managed/cdac/Microsoft.Diagnostics.DataContractReader.Contracts/Contracts/GC/GC_1.cs	Implements contract helpers for alloc-context skipping and SOH/LOH/POH alignment needed by the walker.
src/native/managed/cdac/Microsoft.Diagnostics.DataContractReader.Abstractions/Contracts/IGC.cs	Extends the IGC contract interface with two new helper methods used by heap walking.
src/coreclr/debug/daccess/dacimpl.h	Updates DacHeapWalker private method naming to match the new “advance to next valid object” behavior.
src/coreclr/debug/daccess/dacdbiimpl.cpp	Reworks heap-walk advancement logic to skip corrupt segments cleanly (while surfacing partial-heap) and fixes global alloc context counting.
docs/design/datacontracts/GC.md	Documents the new IGC APIs and the intended algorithms for stepping/alignment.

Copilot's findings

• Files reviewed: 6/6 changed files
• Comments generated: 8

[Copilot]

Copilot's findings

• Files reviewed: 6/6 changed files
• Comments generated: 3

[Copilot]

Copilot's findings

• Files reviewed: 6/6 changed files
• Comments generated: 5

[max-charlamb]

It'd be good to add at least a single dump test verifying that we can successfully get the Heap items.

[rcj1]

It'd be good to add at least a single dump test verifying that we can successfully get the Heap items.

Sure, we can at least verify a handful of items. A more thorough validation against the legacy with all the heap items is done in the internal tests. This test won’t really be useful for diagnosing partial failure, as we can’t know a priori everything on the heap, but if it completely fails then it would be a signal.

[Copilot]

Copilot's findings

• Files reviewed: 9/9 changed files
• Comments generated: 6

[Copilot]

Copilot's findings

• Files reviewed: 9/9 changed files
• Comments generated: 1

[rcj1]

/ba-g timeout

[noahfalk]

@janvorli - fyi for contracts relating to GC