9596; fix M5 performance; check for llama

[xhochy]

We need the newer SDK to support TensorOps on the M5 chips. This makes a massive difference in performance for me. I was getting the following warning:

0.00.051.117 E ggml_metal_library_init_from_source: error compiling source
0.00.051.123 W ggml_metal_device_init: - the tensor API is not supported in this environment - disabling

If we build with a newer SDK, this vanishes and performance increases. E.g. for llama-server -hf unsloth/Qwen3.6-27B-MTP-GGUF:Q4_K_M --spec-type draft-mtp --spec-draft-n-max 2 we go from 220t/s (prefill) and 24t/s (decode) to 550t/s (prefill) and 27t/s (decode).

[xhochy]

I took also the liberty and increased the likelihood of getting new builds as it seemed that recently the bot didn't issue any PR.

[conda-forge-admin]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/recipe.yaml) and found it was in an excellent condition.

I do have some suggestions for making it better though...

For recipe/recipe.yaml:

• ℹ️ The recipe is not parsable by parser conda-recipe-manager. The recipe can only be automatically migrated to the new v1 format if it is parseable by conda-recipe-manager.

_{This message was generated by GitHub Actions workflow run https://github.com/conda-forge/conda-forge-webservices/actions/runs/27368233897. Examine the logs at this URL for more detail.}

[conda-forge-admin]

Hi! This is the friendly conda-forge automerge bot!

I considered the following status checks when analyzing this PR:

• linter: passed
• azure: passed
• github-actions: passed

Thus the PR was passing and merged! Have a great day!