[ggml] Fix Vulkan-Hpp handle usage on 32-bit targets.

[miyanyan]

Overview

On 32-bit platforms, Vulkan non-dispatchable handles such as VkBuffer are represented as uint64_t, and Vulkan-Hpp disables implicit conversions for type safety. This exposes two issues in ggml-vulkan:

1. vk::Buffer is streamed directly into std::ostream in debug/memory logs.
2. vk::Buffer is cast to VkBuffer before being passed to Vulkan-Hpp CommandBuffer::copyBuffer APIs.

Fix these by explicitly casting vk::Buffer to VkBuffer only for logging, and by passing vk::Buffer directly to Vulkan-Hpp copyBuffer calls.

Additional information

Requirements

• I have read and agree with the contributing guidelines
• AI usage disclosure: YES, first used AI to update patch in ggml update to 0.11.0， llama-cpp update to 9030 microsoft/vcpkg#51551, and all ci passed

[miyanyan]

the ubuntu-24-webgpu ci error:

98% tests passed, 1 tests failed out of 44

Label Time Summary:
main    = 994.86 sec*proc (43 tests)

Total Test time (real) = 995.40 sec

The following tests FAILED:
	 39 - test-backend-ops (Timeout)                        main
Errors while running CTest
Output from these tests are in: /home/runner/work/llama.cpp/llama.cpp/build/Testing/Temporary/LastTest.log
Use "--rerun-failed --output-on-failure" to re-run the failed cases verbosely.
Error: Process completed with exit code 8.