docs: Add instructions to install llama.cpp from conda-forge

[jjerphan]

Overview

I would like to let you know that we, a few contributors to conda-forge (notably @traversaro) and I, currently distribute llama.cpp for Linux, macOS, Windows for arm64, x86_64, Apple Metal, CUDA, via conda-forge/llama.cpp-feedstock.

This makes installing and using all of llama.cpp very easy within a few seconds with conda, mamba or pixi.

Additional information

Requirements

• I have read and agree with the contributing guidelines
• AI usage disclosure: NO

[jjerphan]

Hello @ngxson,

Thank you for your work on llama.cpp! What do you think of mentioning this installation option? :)

[ngxson]

@jjerphan just to verify, the build is pulled from the official release, right?

a bit unrelated to this PR: for each distribution platform, I think it worth adding a line somewhere to explain where the build comes from; we should add this via a dedicated PR

[jjerphan]

For all those distributions, the builds come from conda-forge which rebuilds everything from the sources upstream transparently on CI runners like all the other projects conda-forge distributes.

Please let me know which elements you would like indicate so that I can word it specifically.

[jjerphan]

Thank you!

[dev-tinker]

Minor discrepancy noticed: Package name "llama-cpp" actual name is "llama.cpp" in conda-forge.

[jjerphan]

Thank you for this mention! Corrected with #26229.