ggml-backend-meta: fix device and buffer type names

[Frozenlock]

Overview

Fixes the names of meta devices and meta buffer types: the constructors iterate the moved-from constructor parameter instead of the member of the same name, so every meta device and buffer type is named "Meta()".

With a single meta device this is cosmetic (the current -sm tensor path), but any code that distinguishes buffer types by name breaks as soon as more than one meta device exists — e.g. the KV cache context map in llama-kv-cache.cpp collapses the buffer types of distinct meta devices into one entry, allocating all layers' caches on the first device.

Verified: with two meta devices, names render as Meta(CUDA0,CUDA1) / Meta(CUDA2,CUDA3) and the per-layer buffer-type maps no longer collapse.

Additional information

Found while experimenting with tensor parallelism for MLA architectures on a 12x RTX 3090 rig (multiple meta devices in one process). Related follow-up work will be proposed separately.

Requirements

• I have read and agree with the contributing guidelines[1]
• AI usage disclosure: YES. the bug was found, the fix authored, and the validation runs executed by an AI coding agent operating my hardware under my direction.

[1] I am NOT a cpp programmer, so I'm not in a position to evaluate the quality as I would with my main programming languages. That's my way on contributing back to this project. Given the contributing guidelines, I'll understand if you don't accept the PR.

---

Edit:

Results from the follow-up work (for context)

On 12x RTX 3090 (PCIe Gen4), GLM-5.2 744B-A40B at UD-IQ2_M:

config	decode (novel text)	decode (code rewrite)	context
-sm layer (baseline)	23 t/s	23 t/s	384K
MLA TP, hybrid 2x6 + spec	33 t/s	69 t/s	96K
MLA TP, hybrid 4x3 + spec	42 t/s	107 t/s	48K

"spec" = --spec-type ngram-map-k4v,draft-mtp (the GLM MTP head, ported from #24868 to
GLM-DSA). TP output verified byte-identical to -sm layer on GLM-4.7-Flash before any
speculative decoding was added.

Branches:
mla-tensor-parallel
hybrid-tp-pp
mla-mtp

[ggml-gh-bot]

Hi @Frozenlock, thanks for your contribution!

Per our contribution guidelines, the automated PR checker found the following issue(s) that need your attention:

• Multiple open PRs from a new contributor: We limit new contributors (those without a previously merged PR) to 1 open PR at a time. You currently have 2 open PRs.

• AI-generated content: This project does not accept PRs, descriptions or commit messages that are fully or predominantly AI-generated. If you have used AI to assist you in writing code, please make sure to disclose that explicitly.

---

Please note that maintainers reserve the right to make final decisions on PRs. If you believe there is a mistake, please comment below.