It seems like Julia packages, modules, and environments (REPL, IDE, Jupyter, command line) lead to a wide combinatorial proliferation of workflows. I have yet to settle on an efficient one.
The structural constraints of MATLAB, Fortran, or python, while indeed limitations, do lower the learning curve and time to productivity for new users.
Good workflows have to be learned. They can be taught, but not cloned from GitHub. Dissemination of really good workflows is a key thing that would grow the Julia community.