[AndreasKunar]

• I have read the contributing guidelines
• Self-reported review complexity:
  • [x ] Low
  • Medium
  • High

Improvements/Issues addressed:

1. Add documentation on how to build for Windows, especially for ARM (with MSVC or clang)

• See - llama.cpp/docs/build.md

2. Fix, that building with MSVC breaks for Snapdragn X / Windows because MSVC does not support any C in-line Assembly for ARM (_asm_ directive) - fixes Bug: ggml-aarch64.c does not compile on Windows ARM64 with MSVC #8446

• append "&& ! ((defined(_MSC_VER)) && ! defined(clang))" before each _asm_. Clang on Windows masquerades as MSCV, therefore the strange conditional - that it must not be a MSVC which is not clang.

ToDo in future fixes for Snapdragon X on Windows and not addressed by this PR:

• Work on enabling GPU acceleration for SnapDragon X, e.g. via getting Vulkan to run on Snapdragon x - see Bug: llama.cpp with Vulkan not running on Snapdragon X + Windows (Copilot+PCs) #8455

• Work on enabling NPU acceleration for SnapDragon X, e.g. via work on getting QNN to run, and also on Windows - see ggml-qnn: add Qualcomm QNN(Qualcomm Neural Network,aka Qualcomm AI Engine Direct) backend #6869 (currently only for Android)

P.S: This is my first PR ever, and with the help of chatGPT as tutor. Please have mercy on me, if I misunderstood things.