fix(container): update image ghcr.io/ggml-org/llama.cpp (server-cuda-b10200 ➔ server-cuda-b10223)

[drag0n141-bot]

This PR contains the following updates:

Package	Update	Change
ghcr.io/ggml-org/llama.cpp	patch	server-cuda-b10200 → server-cuda-b10223

---

Configuration

📅 Schedule: (in timezone Europe/Berlin)

• Branch creation
  • At any time (no schedule defined)
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about these updates again.

---

• If you want to rebase/retry this PR, check this box

---

This PR has been generated by Mend Renovate.

[claude]

ghcr.io/ggml-org/llama.cpp: server-cuda-b10200 → server-cuda-b10223

Verdict: Safe to merge

Patch bump spanning 23 upstream commits (build b10200 → b10223). Reviewed the full commit range between the corresponding tags:

• Most changes are backend-specific and don't affect the CUDA build used here: SYCL flash-attention (MKL GEMM path), Vulkan (POOL_1D op, driver checks, topk_moe fusion), WebGPU flash-attn, and ZenDNN mul_mat changes.
• The one CUDA-specific change (cuda: extract Q2_0 elements via __byte_perm) is a perf optimization for Q2_0-quantized models; this deployment uses Q4_K_M, so it's not exercised.
• DeepSeek V4 / MTP-related changes (KV cache type enforcement, MTP tensor loading, speculative-decoding token replay fix) don't apply — this repo runs Qwen3.5-4B with no draft model configured.
• Two mtmd (multimodal) commits landed: add n_embd_head (prep work extracted from an in-progress PR, additive only) and add minicpmv46 downsample (MiniCPM-specific). The n_embd_head PR's own test matrix shows the same pre-existing model failures as master (unrelated to Qwen), so no regression expected for the Qwen3.5-4B + mmproj setup in docker/nas01/llama/docker-compose.yaml.
• No changes to server CLI flags/env vars (LLAMA_ARG_*) used in this compose file.

No breaking changes, deprecations, or actionable new features identified for this repo's usage.

Sources consulted:

• ggml-org/llama.cpp@5f55650...11924d4
• ggml-org/llama.cpp#26342
• ggml-org/llama.cpp#25603