Very strange python behavior

I have some python code which is valid python syntax, but fails to run:
```python
a = [1, 2, 3]
b = [3, 2, 1]
c = [a[i] * b[i] for i in range(len(a))]
print(c)
```
Which produces the following error: `Error: NameError: name 'a' is not defined`
This code runs fine on my local python interpreter, but not when I run it through this function:
```rust
pub fn run_python_code(code: &str, inputs: &[(String, String)]) -> String {
let mut config = pyembed::OxidizedPythonInterpreterConfig::default();
config.interpreter_config.filesystem_encoding = Some("utf-8".to_string());
config.set_missing_path_configuration = false;
config.argv = Some(vec!["python".into()]);
config.interpreter_config.executable = Some("python".into());

pyembed::MainPythonInterpreter::new(config)
.unwrap()
.with_gil(|py| {
println!("{}", py.version());
py.run(
"import sys; from io import StringIO; sys.stdout = _stdout = StringIO()",
None,
None,
)
.unwrap();
let error = if let Err(e) = py.run(code, None, Some(inputs.to_vec().into_py_dict(py))) {
format!("Error: {}", e.to_string())
} else {
"".to_string()
};
let output: String = py
.eval("sys.stdout.getvalue()", None, None)
.unwrap()
.extract()
.unwrap_or_default();

py.run("sys.stdout = sys.__stdout__", None, None).unwrap();
let output = output.trim().to_string().replace("None", "");
if error.is_empty() {
output
} else {
error
}
})
}
```
Notably this runs through that function fine:
```python
a = [1, 2, 3]
b = [3, 2, 1]
c = [a * b for a, b in zip(a, b)]
print(c)
```
which should be logically equivalent code
Am I setting up the interpreter wrong or something?

Start by reproducing the first Python snippet through the shown run_python_code function, then compare its py.run invocation and namespaces with the working zip example. Trace the embedded interpreter behavior around list comprehensions and verify the fix by running both snippets and confirming the expected output.

Written by the indexing model from the issue text.