v3.0.1

Final memory system has been released. The docs\ folder now contains router training proposal if you would like to train yourself, but I will release my own trained version using the RLHF pipeline alongside the custom dataset. This release is the canon that supersedes release 1. Version one was pushed to establish baseline so that when I released this update their is gradual upscale of updates not huge one off monoliths. The documents and repository will be cleaned up but as is neural_router.py and memory_injection_system.py are in their final stages. They can work as standalone modules but their true capabilities shine together. This is not a "clone and play" repository or system, these are infrastructure that expect a substrate and an output to be wired in. This is following the philosophy of Operation SOTA. This release however should not be just leveling the oss playing field, this is "score keeper. Plus one" against the SaaS empire. This revision includes new XML formatted persistent memories, a matured json import, semantic retrieval on demand(cross chat referencing), alongside tiered capabilities being unlocked as that very notion stands against the philosophy.