Optionally Use Fickle for Deserialization of Data Returned from Untrusted Code

[yeldarby]

Describe your changes

Adds optional integration with rffickle for safe pickle deserialization to prevent Remote Code Execution (RCE) attacks when running untrusted user code in Modal sandboxes.

• Adds use_firewall parameter to @app.function() and @app.cls() decorators
• When enabled, uses rffickle.DefaultFirewall to block dangerous pickle operations during deserialization
• Per-function configuration allows selective protection for untrusted code while maintaining full pickle support for trusted functions
• No unsafe fallback - if firewall is requested but rffickle is unavailable, the function fails rather than falling back to unsafe deserialization

Compatibility checklist

• Client+Server: this change is compatible with old servers
• Client forward compatibility: this change ensures client can accept data intended for later versions of itself

Release checklist

• Version file (modal_version/__init__.py) has been updated with the next logical version
• Changelog has been cleaned up and given an appropriate subhead

Changelog

• Added use_firewall parameter to functions and classes for safe deserialization of untrusted pickle data using rffickle
  • Enable with @app.function(use_firewall=True) for functions that process untrusted code
  • Also supports Cls.from_name() and Function.from_name() with use_firewall=True parameter