[Feature Request] Interested in developing Execution Provider using MLIR/LLVM - Is There Existing Work or Execution Provider based on MLIR or LLVM?

### Describe the feature request

I'm interested in developing a CPU Execution Provider that leverages MLIR and LLVM infrastructure. Before I begin, I’d like to know if there’s any existing work, roadmap item, or ongoing effort in this area.

If not, I’d be happy to start exploring this and contribute to the project. Any guidance on approaching this within the current architecture would be greatly appreciated.

### Describe scenario use case

For example, RISC-V CPUs currently lack a community-maintained Execution Provider. Using MLIR and LLVM to build one would allow us to leverage compiler-driven automatic optimization like vectorization and target-specific code generation without requiring manual tuning. This makes it easier to support new or evolving CPU architectures efficiently

Start by reviewing ONNX Runtime's current execution-provider architecture and checking for existing MLIR or LLVM efforts. Clarify whether a roadmap or implementation already exists, then define the scope and integration approach for a RISC-V CPU execution provider; the issue provides no specific files or tests.

Written by the indexing model from the issue text.