feat: add llmman sample as a local model provider

[ericcurtin]

Adds llmman, a local model runner that serves Ollama-, OpenAI- and Anthropic-compatible APIs on port 17434.

• This repo integrates Ollama in the examples via its OpenAI-compatible /v1 endpoint rather than a provider package, so llmman is added the same way.
• examples/LangChain.Samples.Llmman: minimal chat sample modelled on LangChain.Samples.OpenAI, pointing OpenAIClient at http://127.0.0.1:17434/v1 (model gemma4). Honours LLMMAN_HOST ([host][:port]) like OLLAMA_HOST; uses a placeholder API key since llmman needs none.
• LangChain.slnx: registers the sample under /Samples/ (alphabetical).
• examples/LangChain.Samples.LocalRAG/Program.cs: one-line comment that llmman serves the same API on port 17434.

Testing: .NET SDK 10.0.400: dotnet build of the new sample and LangChain.Samples.LocalRAG (0 warnings, 0 errors), dotnet format --verify-no-changes on the new project, dotnet sln LangChain.slnx list resolves it. Not run against a live llmman server.

AI-assisted, reviewed before submitting.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}