Update LLamaEmbedder, Examples packages, and KernelMemory examples

- Embedding generation: Extension with Batch processing + Normalization (important to have this built-in for KernelMemory).
- Examples had wrong nuget packages, updated to correct ones.
- Updated KernelMemory examples.
- added missing model parameters
- adding config is null check
- unit tests project update to prevent the constant download of many GBs
- ** for some reason Embeddings must be set to false in the kernel memory text embedding generator => we need to follow this and check it later because this should normally be 'true' ! **
- skipping one test for macOS (all other tests are OK)
- setting GpuLayerCount to 0 in Release in CIGpuLayerCount also for Windows
- possible BUG in llama.cpp in 'if (params.split_mode == LLAMA_SPLIT_MODE_NONE)'... trying to set other split mode (even if there is no GPU)!

Author: zsogitbe

LLama.Unittest/KernelMemory/ITextTokenizerTests.cs

@@ -22,7 +22,7 @@ public ITextTokenizerTests(ITestOutputHelper testOutputHelper)
             _testOutputHelper = testOutputHelper;
 
             _infParams = new() { AntiPrompts = ["\n\n"] };
-            _lsConfig = new(Constants.GenerativeModelPath) { DefaultInferenceParams = _infParams, ContextSize = 512 };
+            _lsConfig = new(Constants.GenerativeModelPath) { DefaultInferenceParams = _infParams, ContextSize = 512, SplitMode = LLama.Native.GPUSplitMode.Layer };
 
             testOutputHelper.WriteLine($"Using model {Path.GetFileName(_lsConfig.ModelPath)}");
         }