Sounds more acceptable.
Maybe is good idea to cooperate in this matter in near future.
It seems that there could be a common belief that GPL licensed software might be part of some sort of religious anti-commercial cult, but I don't think it should necessarily be the case, especially in projects like this. The reason why I added "special permission" exception (although it can be implied in any case anyway) is that original author, who owns full copyright (that excludes any third-party contributed code, of course), is not bound by GPL and can actually re-license the product or grant permissions on it as needed.
So, in other words, as long as you want to use open-source project for academic and/or scientific purposes and don't really care about GPL, you don't ever need to think about it. On the other hand, you can still get a special permission to use the project under different terms and therefore might not even bound by GPL at all, depending on the permission given. This is why I think it is pretty flexible.
The only limitation comes for contributions, which have to be bound by GPL and likely to remain this way, but given current situation, in Pascal community, there are relatively few of us who develop frameworks. I felt this myself when had to find people to hire for one project a couple of years ago - finding someone with knowledge and experience in Object Pascal was *extremely difficult*, if not semi-impossible. If the situation changes and community grows, with more activity and contributions, I can always re-release the framework under more liberal license to put less burden on contributions.
I don't know what is more efficient: approach presented by you or more mORMotish triple licence approach.
After re-opening support forums for the project, I was quite surprised by the small amount of traffic, unlike some time ago like before 2010. Around that time and earlier, even some time before, when Asphyre project was Delphi-only, there was a huge traffic and many people interested in DirectX/OpenGL. Relatively recently, I have been watching posts here on forums along with some others regarding similar topic and really there are only a couple of people into this kind of technology, and unfortunately, for the most part, only those people who themselves work on frameworks, with very few, if any, being actual users. Maybe there is a huge hidden Pascal user base that I don't know about, but so far I couldn't find it. Therefore, I think that my current, sort of personalized approach, is efficient enough. Again, if situation changes, we'll see...
For a contrast, if a similar library would be written in, say, Python, there would likely be many more people interested in contributing, maybe even some big companies, which would definitely benefit better from a more liberal license terms, where you don't have to contact the author.
...GPL projects (especially libraries) do not exist for me. They are like transparent ghost, outside my interest. IMO is better for you to release PXL as full commercial library instead of GPL.
As I've said above, perhaps it's some sort of cult present in some GPL projects and/or developers that may give this overall perception. I didn't choose GPL as means to restrict PXL usage, but merely as a means of good legal protection for my investment - after all, I use PXL myself (and it was actually revived and developed for that specific purpose) for scientific work; could have kept it to myself. Having said that, I am not against its usage in commercial/closed-source projects, which is why I've provided the alternative to contact me for an arrangement.