Fix/ci build system

[M4X1K02]

Hey :)
I've noticed the CI failing, so I've prompted gemini how to fix it. This is what we propose:

PR Description

Overview

Resolves CI failures on master by separating the core library build from tool-specific dependencies. Also updates the build system and improves Nix support.
This should make the CI workflow less dependent, as simapi now builds without simd initially.

Detailed Changes

1. Build System (CMake)

Root CMakeLists.txt

• Updated cmake_minimum_required to 3.15.
• Added BUILD_SIMD option (default ON) to allow building without simd deps.
• Made simapi include directories PUBLIC.

simd/CMakeLists.txt

• Fixed compiler flags previously targeting a non-existent monocoque target.
• Improved argtable search logic for portability.

2. CI & Infrastructure

GitHub Actions

• Core library job now uses -DBUILD_SIMD=OFF.
• Nix-based job remains for full simd builds.

Nix (flake.nix)

• Removed self.submodules = true to avoid failures due to missing codegen/cfile.
• Switched to correct argtable attribute.
• Separated nativeBuildInputs and buildInputs.

Verification Results

• Local: cmake -B build -DBUILD_SIMD=OFF builds successfully.
• Nix: nix build .#simapi and nix build .#simd both succeed.
• CI: Addresses the Neither argtable3 nor argtable2 found failure observed previously.

[Spacefreak18]

there's still some issues with the CI though.

for one, it's building a simd that's linked against argtable3, when most distros are still using argtable2.

[M4X1K02]

that is true, I didn't notice that 👍
i will see what i can do.

btw, i would also like to use
cmake --build build instead of

mkdir build
cmake .. etc

this would create the build directory automatically, while also being generator-agnostic!
i would also need to update the monocoque install script + readme though...