programming generic state machines (that would handle a custom data science App): regular .m code vs Simulink / Stateflow?

Usually, one would use Simulink/Stateflow for HDL/Bio/Automotive/Embedded-FPGA/C++/etc. scope ... which a highly focused application with immediate benefit of direct conversion into c code (for example).

Nonetheless, why would the average data cientist Joe not use Stateflow while working on below such mundane tasks:

1) large quantity of text data processing (curating, adjusting, alignment, etc)

2) incorporation of such data (which can be financial timeseries for example) into a custom workflow

3) state machine implementation (involving, not a huge number of state, but say - something such as HFT)

Not being a Simulink user, I'm just curious why I haven't seen peepl in such fields as above, using Simulink in their day to day endeavors.

More specifically, what are the pros, what are the cons? Up to where it's overkill, beyond which point complexity demands usage of Stateflow for such use-cases?