The classes in the U.S. are definitely separated by name, but the content of the 4 "Calculus" courses I took for electrical engineering wasn't quite as distinct as some are portraying it. We spent a substantial amount of time on sequences/series, limits, convergence, Taylor series, differentiability and integrability, etc., and related theorems. It was quite a bit more than "how to compute"; however, it wasn't nearly as rigorous as the proof-based real analysis class my son described from Caltech.
I'm sure that the engineering disciplines are a big reason for the distinction. The distinction may have other historical roots, but engineering schools sustain it. In the U.S., engineering (especially electrical) has changed in the last 50 years because the world is changing; instead of Fortran, it's incredibly helpful to know C++, Python, AND MATLAB; Newer high-speed transistors has shifted emphasis to RF circuit design classes; Increases in computer frequencies is pushing signal/power integrity to the forefront because-- in industry-- it's a dominant issue that barely existed in the 1980s. One EE prof. I spoke with said that electromagnetics, which formerly included distinct classes in electrostatics & magnetics, was condensed into 1 course that is barely adequate. There is an ever-widening list of courses/requirements and many students can barely finish a degree in 4 years; Math is just one of several areas receiving pressure to shed "unnecessary" requirements. My point is that real analysis would be good for engineers, but it's competing with many other math/non-math topics.
One can argue whether engineering should be a 5 or 6 year degree; but unless someone forces that upon everyone, which university will be the first to step forward and inform students that they must now pay for 5-6 years of school? In my opinion, economics is the sustained pressure to streamline courses and focus on practical "calculus" topics.