I think you get more benefit from reading code if you study something very close to what you are working on yourself, something in the same domain, in the same framework perhaps, or at least in the same programming language, at best something you are deeply involved in currently.
I never seem to get enough motivation to read deeply into random "grand" code bases like Lua or SQLLite, but some months ago I got into the habit of always studying a bunch of projects that use a given technology before I use this technology, and it greatly decreased the amount of time it takes me to get to a "idiomatic" coding style. So instead of diving in a random, I would recommend making researching existing code-bases related to what you are currently doing an integral part of your workflow.
I really enjoy doing this as well. Are you aware of any resources or metrics (number of contributors on github, etc.) to find which projects are "well crafted" in X framework/language?
Perhaps I overestimate how much risk there is in learning idiomatic practices from a project which is not actually all that idiomatic. I like the assurance that what I read is quality, especially with frameworks or languages I'm very new to where it can be tough to tell.
I never seem to get enough motivation to read deeply into random "grand" code bases like Lua or SQLLite, but some months ago I got into the habit of always studying a bunch of projects that use a given technology before I use this technology, and it greatly decreased the amount of time it takes me to get to a "idiomatic" coding style. So instead of diving in a random, I would recommend making researching existing code-bases related to what you are currently doing an integral part of your workflow.