First, if one is familiar with a language or invested in a language, he will be biased to choose that language and it makes sense for him. Before you would argue, I admit I can't prove that you are biased, but nor could you prove you are not biased. So let's agree that some anecdotes like your case is not a strong argument.
So let's look at it in another angle. If physicist/people choose FORTRAN not merely due to apprenticeship, it necessarily means that there will be people choose FORTRAN even when his master do not introduce/require him to use FORTRAN. In another word, we need examples of people who choose FORTRAN despite of non-familiarity to FORTRAN. There are plenty of such cases for, e.g. rust. For me at least, I know none such. How about you? For example, how successful have you been influence other people (other than your student) to use FORTRAN? On the other hand, I do know people who choose other language despite of the familiarity of FORTRAN.
The prerequisite for using Fortran is that someone works in a field that relies heavily on large scale, high performance numerical computation. Since Fortran is pervasive in that domain, it's extremely likely the they would be exposed to Fortran through "apprenticeship". Basically, most people who should be using Fortran are using Fortran! I've seen a few instances where people used C/C++ or Matlab in situations where Fortran would have been more appropriate. Matlab has its own drawbacks, but it is also a very domain-specific numerical language, and likewise its users are exposed to it through apprenticeship! C/C++ is still the default language for high performance computing outside of numerics, and thus it's not surprising that people who have not been exposed to Fortran would attempt to use it for numerical purposes as well.
There isn't exactly a lot of choice for numerical programming: Python and Matlab are (partly) designed with numerical purposes in mind, but don't scale to high performance. C/C++ are performant, but not expressive for numerics. There really isn't any other language that would even be in consideration, besides now Julia, but Julia is extremely new, so time will tell.
I don't know anything about Rust, but based on reading its Wikipedia article its not a language designed for numerics, but basically something designed to replace C++ for systems programming. Thus, like C/C++, it will be performant, but not expressive for numerical purposes.
If Fortran is good for computations, and it's fast, and it does what it is designed to do, and does it well, why should it die out?
As a computer scientist and engineer, I do not understand the animosity towards Fortran?
C replaces FORTRAN. That is from merit point of view. In reality, computational physicists mostly goes to C++ plus Python. Truth to be told, physicists are not computer scientists and they are not even into it. They simply picked out of necessity/convenience -- which was the reason FORTRAN was picked then.
PS: That FORTRAN is fast is over stated IMHO. I am familiar with high performance computing. In terms of parallel computing, it all boils down to algorithm design. In terms of vector/linear algebra, it all depends on machine tuned BLAS (which is mostly due to hand crafting rather than the language itself). There is no evidence FORTRAN code runs necessarily faster than C -- comparing first craft non-optimized code. Unlike C, FORTRAN is less capable and more cumbersome to do trendy stuff -- this is the actual merit of FORTRAN over C.
From my point of view, it's understated, and should be given much more credit than it gets. intel has done a lot of work to make their ifort compiler leverage their latest instruction sets like AVX2, resulting in even bigger speedups, and PGI has done absolutely amazing work of writing a compiler which can ingest unmodified Fortran code and compile it to run on NVidia GPU's. It's nothing short of mind blowing to me and gets way too little mention. Modern Fortran is a really nice language and all this bashing of it because it's not trendy is completely unfounded. Just ask intel.
However, there is slight change. The new generation of masters often do know other languages, so they sometime tolerate their new apprentices to use other language and still able to teach them. So at this point, by far, not all physicists use FORTRAN.
FORTRAN will die out, just takes time.
One of the sticking point is the library BLAS/LAPACK. It is in FORTRAN. Or more precisely, its interface is in FORTRAN. I don't foresee this will change for a long time. So FORTRAN will be like COBOL, it will persist even when no-one really use it anymore.