0.116.0

Fixed

• memo search all and meaning search over notes and files answer in a fraction of a second (cli-core 0.451.0).
  Ranking documents and notes by meaning now uses the vector index instead of reading every chunk in the
  store: on one archive with 600,000 chunks, mostly mail, it took 7–18 s a query and now about 0.05 s, with the same
  results. It still reads every chunk when the index is not ready.
• Meaning search runs on the native ONNX Runtime on Windows too. After memo models download e5-small, the model
  uses the native library on Windows x64 and arm64, as it already did on Linux and macOS: 2–3× faster than
  WebAssembly. The download takes onnxruntime.dll from the onnxruntime-node npm package (113.5 MB, once, checked
  by sha256).