Python has had for a very long time both NASA and JPL as marquee customers/users, eg. http://montepy.jpl.nasa.gov/ for mission planning and control applications.
From the MONTE brochure:
The story of MONTE begins in 1998, when JPL’s navigation section commissioned an update to the aging DPTRAJ/ODP library. The goal was to translate these time-tested navigation algorithms from Fortran into a more maintainable, extensible, and better tested C++/Python application.
At the end of the day, if Python/C++ is effective for their team and works, then it gets the job done. Personally, I just dislike writing math/simulation code in Python. The syntax is too limiting for that topic and I've run into rather annoying delays when parsing CSV data due to utf8 vs byte vs ascii handling differences when mixing Pyhton 2/3. But mainly, deploying Python code to other users is a pain.
From the MONTE brochure:
The story of MONTE begins in 1998, when JPL’s navigation section commissioned an update to the aging DPTRAJ/ODP library. The goal was to translate these time-tested navigation algorithms from Fortran into a more maintainable, extensible, and better tested C++/Python application.