speculative decoding in llama.cpp

[northcraft90]

Appreciate the detail. The pattern around continuous batching has been pretty similar for me. Worth rechecking your defaults though, because one bad assumption can make the rest of the thread look confusing.