fix(core): reject IPC registration without IPC state

[aryanputta]

Fixes#2568.

DeviceMemoryResource.register() and PinnedMemoryResource.register() can reach MP_register without IPC enabled. The current implementation dereferences self._ipc_data, causing a process-level segmentation fault instead of a Python exception.

This change:

• raises RuntimeError("Memory resource is not IPC-enabled") before the dereference;
• preserves the existing-resource fast path;
• adds coverage for both device and pinned memory resources;
• adds a release note.

Validation:

• Ruff check and format passed.
• Cython lint passed.
• Focused pytest collection was blocked on macOS because the CUDA test helper requires Linux libc.so.6.
• Lychee was blocked by sandbox network access to existing external links.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[lijinf2]

Thanks for the contribution!

This looks like a duplicate of #2569, which was opened by the issue author to address the same bug in #2568. Both PRs raise RuntimeError("Memory resource is not IPC-enabled") when register() is called on a non-IPC memory resource.