Fix dotnet publish // LLamaSharp conflicts (#500)

## Motivation and Context (Why the change? What's the scenario?)

LLamaSharp doesn't allow to include multiple backends at once in a
packaged .NET app, requiring to build once for each backend (CPU,
CUDA11, CUDA12, etc). The current code fails to "publish" due to
concurring backends.

See also #498

## High level description (Approach, Design)

Solve the "publish" issue by removing all LLamaSharp backends, leaving
it to the end user/developer to choose one.

The need to build multiple times, once per LLamaSharp backend, persists,
due to how LLamaSharp is designed. (if anyone knows of a workaround
please let us know!)

Directory.Packages.props

@@ -15,7 +15,9 @@
     <PackageVersion Include="HtmlAgilityPack" Version="1.11.61" />
     <PackageVersion Include="LLamaSharp" Version="0.12.0" />
     <PackageVersion Include="LLamaSharp.Backend.Cpu" Version="0.12.0" />
+    <PackageVersion Include="LLamaSharp.Backend.Cuda11" Version="0.12.0" />
     <PackageVersion Include="LLamaSharp.Backend.Cuda12" Version="0.12.0" />
+    <PackageVersion Include="LLamaSharp.Backend.OpenCL" Version="0.12.0" />
     <PackageVersion Include="Microsoft.ApplicationInsights.AspNetCore" Version="2.22.0" />
     <PackageVersion Include="Microsoft.Bcl.AsyncInterfaces" Version="8.0.0" />
     <PackageVersion Include="Microsoft.Data.SqlClient" Version="5.2.0" />

examples/105-dotnet-serverless-llamasharp/105-dotnet-serverless-llamasharp.csproj

@@ -5,6 +5,14 @@
         <ImplicitUsings>enable</ImplicitUsings>
     </PropertyGroup>
 
+    <ItemGroup>
+        <!-- To run the example choose one of these backends -->
+        <PackageReference Include="LLamaSharp.Backend.Cpu" />
+        <!-- <PackageReference Include="LLamaSharp.Backend.Cuda11" />   -->
+        <!-- <PackageReference Include="LLamaSharp.Backend.Cuda12" />   -->
+        <!-- <PackageReference Include="LLamaSharp.Backend.OpenCL" />   -->
+    </ItemGroup>
+
     <ItemGroup>
         <ProjectReference Include="..\..\extensions\AzureOpenAI\AzureOpenAI.csproj" />
         <ProjectReference Include="..\..\extensions\OpenAI\OpenAI.csproj" />

examples/105-dotnet-serverless-llamasharp/Program.cs

@@ -3,6 +3,19 @@
 using Microsoft.KernelMemory;
 using Microsoft.KernelMemory.AI.OpenAI;
 
+/* PLEASE NOTE:
+ *
+ * The example is pre-configured to use a CPU backend.
+ *
+ * If your device provides CUDA or other backends, you can configure the example by editing the project file (*.csproj)
+ * and editing the LLamaSharp PackageReference, removing the CPU backend and choosing a different one.
+ *
+ * Kernel Memory web service is configured to not have ANY pre-configured LLamaSharp backend, e.g. KM Docker
+ * image does not work with LLamaSharp without manually mounting a LLamaSharp backend assembly into the
+ * Docker image at runtime. Same for custom .NET apps, which need to choose a LLamaSharp backend manually
+ * installing one of the available LLamaSharp.Backend packages.
+ */
+
 var llamaConfig = new LlamaSharpConfig();
 var azureOpenAIEmbeddingConfig = new AzureOpenAIConfig();
 

extensions/LlamaSharp/LlamaSharp/LlamaSharp.csproj

@@ -14,8 +14,16 @@
 
     <ItemGroup>
         <PackageReference Include="LLamaSharp" />
-        <PackageReference Include="LLamaSharp.Backend.Cpu" />
-        <PackageReference Include="LLamaSharp.Backend.Cuda12" />
+
+        <!-- Kernel Memory web service is configured to not have ANY pre-configured LLamaSharp backend, e.g. KM Docker
+             image does not work with LLamaSharp without manually mounting a LLamaSharp backend assembly into the
+             Docker image at runtime. Same for custom .NET apps, which need to choose a LLamaSharp backend manually
+             installing one of the available LLamaSharp.Backend packages. -->
+
+        <!-- <PackageReference Include="LLamaSharp.Backend.Cpu" />      -->
+        <!-- <PackageReference Include="LLamaSharp.Backend.Cuda11" />   -->
+        <!-- <PackageReference Include="LLamaSharp.Backend.Cuda12" />   -->
+
         <PackageReference Include="System.Linq.Async" />
     </ItemGroup>
 

service/Service/Service.csproj

@@ -20,6 +20,15 @@
         <PackageReference Include="Swashbuckle.AspNetCore" />
     </ItemGroup>
 
+    <ItemGroup>
+        <!-- If you plan on using LLamaSharp, you will need to enable one of these backends,
+             which depend on the hardware hosting Kernel Memory code. -->
+
+        <!-- <PackageReference Include="LLamaSharp.Backend.Cpu" />      -->
+        <!-- <PackageReference Include="LLamaSharp.Backend.Cuda11" />   -->
+        <!-- <PackageReference Include="LLamaSharp.Backend.Cuda12" />   -->
+    </ItemGroup>
+
     <!--
     This dependency is needed only to generate appsetting.development.json via