fix(coderunner): handle null/true/false JSON literals in Python sandbox

[octo-patch]

Fixes #2618

Problem

When a workflow code node parameter has a null value, the Python sandbox crashes with:

NameError: name 'null' is not defined

Root cause: json.dumps() serializes null/None values as the JSON literal null, which is then embedded directly into generated Python code via string interpolation:

code = prefix + f'args={json.dumps(params)}\n' + user_code + suffix
# produces: args={"input": "fdsa", "lists": null}

Python does not recognise null, true, or false as keywords — these are JSON literals but not Python identectives. Python uses None, True, and False instead.

Solution

Add null = None, true = True, false = False aliases to the generated code prefix. This ensures that JSON-serialized parameter dicts evaluate correctly when the code string is executed by Pyodide:

null = None
true = True
false = False

With these definitions in scope, the generated line args={"input": "fdsa", "lists": null} is valid Python and evaluates null as None.

Testing

Manually verified that a code node with {"input": "fdsa", "lists": null} input no longer raises NameError. The aliases are defined before any user code runs, so they do not conflict with user-defined names in the main() function scope.