But yes, a less-bitter version of you might be a good EIR, although I think it usually comes from "founder of a company funded by the firm has a midsize exit", not out of the blue.
Three of my friends were EIRs, and two are now founders of great companies (although not hardcore tech companies). The other has run off to Spain for a while, after ~15 years of doing startups, for something of a sabbatical. They're all at the "code for days straight", graduated from top engineering programs (in contexts like being in India or ~20 years ago where that mattered, unlike the US today), etc.
The whole point of EIR is that it's revolving door -- you do it for a year or two, then go back to startups. It's both a way to take a break, and get exposed to new stuff. You still use some of your tech skills, and often end up working on side projects, so I don't think a year off is going to kill your technical competence.
As for the quality of portfolio company review/advice: one top one sent me to do an interview/etc. with probably the best operating executive in the security industry, one of their portfolio companies. I learned a lot, and decided we should hold off on raising as a result (this was ~2 years ago, about a month after I finished YC).
Being an actual full-time VC probably does hurt your technical skills, and I don't think having just fairly narrow programming or programming tech skills is a useful background for a VC, although "ability to pick winning teams" is essentially the definition of a successful accelerator or seed stage investor (but, on more axes than just "can program well"). Something a bit broader and maybe less deep is good, although in cleantech having postdoc level expertise might make sense. I think I have more than enough security industry experience to be an effective security vc, but not enough in the programming tools space, and somewhere in the middle on deployment/automation/networking tools.
I've talked to VC Principals/Venture Partners who are what I'd consider top-50 domain experts in specific things (payment regulations), and some Partners who are top-500 on networking or certain parts of security.
The other thing is it's usually better to have domain expertise in the founding team than implementation perfection, so at seed, a PHP hacked together piece of crap that works is probably fine, as long as what they're doing 1) has a market 2) is possible. At Series A, they should probably have some competence in the team, and after, sure.