By all metrics that I have ever seen, Pascal/Object Pascal was a top 10 programming language from the late 1970s to the early 2000s. That Pascal/Object Pascal would presently be in the 15th to 20th range among programming language is logically consistent with its past. Contrary to the opposing view, Pascal/Object Pascal is not going to all of the sudden disappear. There is still a ton of code, websites, books, teaching materials, it being used in the schools of numerous countries throughout the world, and Delphi is still in existence and viable.
I don't think pascal will disappear at all, if cobol has proven anything than that technologies never die. The question is what happens with a language that once was very popular, and is this what we want from pascal.
First option is like cobol that the language goes into "maintanance mode", i.e. people only use it to maintain and continue already existing code. This is for example the fate of cobol. Pretty much no one is going to start a big new project in cobol, but due to prevalence it once had, there are a lot of systems that need maintainance, and this is why cobol will probably stick with us for at least the next 20 years. This is the very least we can expect of Pascal, especially as backwards compatibility is a big concern for the development of both Delphi and FreePascal.
The second option is that it becomes a niche language, like Fortran, which is the de-facto standard in the high performance computing area, but does not really find any usage somewhere else. Currently I don't see this for Pascal, because up until now Pascal was a general purpose language, and I don't see any special area where it is exceptional at. There once was the cross plattform GUI development, but as I said, there are now other options available. Maybe for Desktop only GUIs like it is often required for some enterprise applications like management software, but here are also a lot of other players, mainly Java with Swing and (if you stick to windows, which is a reasonable assumption for many companies) .Net. Also this niche is where Delphi sits, and I don't see how Lazarus could compete there with Delphi (enterprise software development companies usually want to have someone to shout at when it doesn't work, which doesn't work as good in an open source project).
Lastly is the possibility of innovation, to keep up with the "new hot stuff", continually evolve and stay relevant in the ever shifting development market. This is what for example C++ does. C++ has gone through many iterations and the language has changed, in many cases fundamentally. C++ code from the early 00s is completely different from modern C++ code, every few years with a new standard, a lot of new things are introduced, some are successful, others don't see much usage and some are even dead on arrival and don't even make it into most compilers. This keeps C++ always modern and is arguably a reason why it is keeping it's place as a very popular language, while C on the other hand is steadily declining into it's niche (of very bare metal development).
Other languages like Python or Swift even go so far to make a complete cut, in 08 or so, the Python comittee decided that python wasn't up-to-date anymore, which is when they decided to create python3 which completely breaks compatibility with the old python2, to basically start from scratch. Swift did something similar in version 3.
So the question is what the goal is. I personally think Pascal and Lazarus are best suited as general purpose languages, so I would preferr the third option, because I really like the language and would love to have it be as widely applicable as possible. Thats why I really like to experiment with new features, modern paradigms and co.
But honestly, I think I am in the minority here, a lot of people seem to be very satisfied with the fact that you can still write code as 20 years ago (which is completely fine, this is a completely valid goal to have) which consequently means that the language needs to go either the option1 or option2 route.