Introduce CMake fetchcontent and CMake Install Workflow

[MathewBensonCode]

FetchContent Package Management

• Here is an idea for migrating from the git module based dependency management to use FetchContent.
• The packages are still built by the build system in order to preserve the versions.
• This is done by avoiding the use of the Find_Package method that would use what is installed on the system.
• This makes it virtually the same.

CMake Install

• This replaces the same functionality that was in the PostBuild.ps1 script
• Used the cmake install features to ensure consistent installation location, which is canonical i.e. binaries in the bin directory and libraries in the lib directory
• As I was debugging I found that there was some inconsistency with the paths to the shaders in Panzerfaust vs Obelisk. e.g. if you ran obelisk directly without going through Panzerfaust the built in shader location wasn't the same.

SwapChain Image Sizes

• Was getting a vulkan error that the system was running out of memory.
• Traced this to the size being passed to the SwapChainCreateFunction.
• Initialize SwapchainImages with the size parameter added as the creation of the swap chains was happening successfully but the size was not getting updated giving an index error.

Wayland on Linux

• Fix error on linux where Wayland was not being initialized properly. Wayland is the default for glfw in version 3.4 and in many desktop environments. GLFW_EXPOSE_NATIVE_* macro was defined for windows but not for the other Operating Systems

[MathewBensonCode]

@JeanPhilippeKernel @jnyfah I'm not sure why this check is failing, however, I would like to request you to try the build artifacts on your systems.

The main thing with the install workflow is to move what was in the Panzerfaust/net8.0..../publish folder is not going to the bin folder with the vulkan loader library in the lib folder.

This means that downloading the release builds should be able to run on a user system easily.

The program doesn't run because of the "Shader::CreateDescriptorLayout()" code runs out of vram when loading up the shaders.

However, I believe that the "not running on mac and linux" issue may be resolved by this.

[JeanPhilippeKernel]

Hey @MathewBenson
Thanks a lot for the PR - I re-ran the pipeline and it's green now -

[JeanPhilippeKernel]

my first observation will be to extract the vulkan related issue in a seperate PR - and as heads-up there is an on-going effort to fix the issue here : #479