How many hours they code is probably a better indicator. Commit times and frequency aren't really indicative of hours invested in coding.
This post needs a disclaimer as "speculation"
I don't agree. Environments exist where people code 12+ hours a day, six days a week. Commit times and frequency are likely influenced by time invested in coding. Could you spend 12 hours on a problem with only one commit? Definitely! But if your commit window is spread across 12+ hours a day and the frequency of commits within that window is non-insignificant, it's unlikely that is artificially expanded from less work. It is a signal worth extracting.
You'd want to cast a net over as much of the engineering talent in an org that you can to find useful trends.
I would pay for this as part of some sort of Glassdoor/Stackoverflow Jobs hybrid. Only show me jobs where work/life balance doesn't suck.
Regardless of my productivity during the week, there are times during weekend where I wanted to continue working on a work project just because I genuinely enjoyed it or because there was something small I could do that would save a lot to the project/company and I did not mind at all.
IMO bringing work into your personal life does not necessarily indicate a poor work/life balance, being invested in your work either.