[3.14] gh-148022: Add threat model to remote debugging docs (GH-148024) (#148039)

gh-148022: Add threat model to remote debugging docs (GH-148024)

The remote debugging protocol has been generating spurious
vulnerability reports from automated scanners that pattern-match
on "remote access" and "memory operations" without understanding
the privilege model. This section documents the security boundaries
so reporters can self-triage before submitting.

The threat model clarifies three points: attaching requires the
same OS-level privileges as GDB (ptrace, task_for_pid, or
SeDebugPrivilege), crashes caused by reading corrupted target
process memory are not security issues, and a compromised target
process is out of scope. A subsection explains when operators
should use PYTHON_DISABLE_REMOTE_DEBUG for defence-in-depth.
(cherry picked from commit edab6860a7d6c49b5d5762e1c094aa0261245a9c)

Co-authored-by: Pablo Galindo Salgado <Pablogsal@gmail.com>