@: How is what you're suggesting any different than the way such validation has always been done?
Client-side language validators (such as the HTML validator W3C provides) have no idea what scripting language you're using, if any. Using PHP to generate an HTML webpage is no different than using a plain .html file or even some new server-side language that hasn't even been invented yet. In the end, the client is never any wiser and simply processes whatever output it receives.
In other words, I don't see how the process you described above is any different than before PHP was even invented; in the end, you've got some HTML/client-side language that you want client's browsers to interpretet correctly. That's it.
In fact, W3C's validator allows you to specify a URL, so there's no need to even save the outputted HTML webpage at all.