[Enhancement]: Batch note embedding for faster bulk indexing

[HahaBill]

Right now notes are embedded one at a time. processNextNote sends a single chunk to the worker and waits for the result before sending the next, so there's only ever one request in flight. This can be slow for a large number of notes and it scales linearly with note count.

Figure out how to do this concurrently/in parallel.

[HahaBill]

Nice performance optimization after the pipeline is implemented and in steady state