maybe the doc should add some instructions at geting started section

the doc surely has these content, but as a newbie it will still be confused.
i think should add instructions on getting started section as follows:

1. input.advance_to: it make dd wont accept input changes older than the time.
2. input.flush: advance_to is just a buffer, flush labeled this will happen.
3. worker.step: the above is just configs, and only work.step will make this happen.
4. when exit dd, dd will call worker.step once, make all the work happens. but when you write your loop or infinite graphs, you should call worker.step whenever you need your computation works
5. in a nut shell, in your program, you should call 1-3 sequencetially, otherwise you maybe see hungs.

i think these will make things easier

Start by locating the Getting Started section in the repository documentation and read the surrounding explanation of input.advance_to, input.flush, and worker.step. Clarify the sequence and shutdown behavior described in the issue, then update that section with the numbered guidance and verify that the instructions match the existing examples.