I'm not sure if this type of entry fits the purpose, since I didn't use bookdown and the output is LaTeX-only at this stage, but the criteria seemed to encourage anything rmarkdown-related. Anyway, since I've already written a blog post about this project I thought I'd show a quick preview here.
I also made a resume with a similar workflow.
These two projects didn't require the knitr step, so the input format is plain markdown + yaml, but changing the extension to Rmd will run any code chunk as usual.
A Lua filter is used to (minimally) process the intermediate Abstract Syntax Tree produced by
pandoc internally; this step is quite interesting as it provides fine-tuning not easily achievable at earlier/later stages in the process, and could potentially be done with R using the
pandocfilters package (I haven't explored this option yet, as the documentation is quite scarce).