Document the non security model of Coq

[Zimmi48]

we should make clear that Coq is not intended (at all) to be used by adversaries. Right now, from the security point of view, if you load in Coq a malicious script it could likely create a lot of mess, as basically Coq has the capability to overwrite almost any file in the system [that the user has access]

I would be fine with closing that gap, or just documenting the (non) security model of Coq.

Originally posted by @ejgallego in #10232 (comment)

[Alizter]

Here is an example for python:
https://python-security.readthedocs.io/security.html