Missing quotation marks in lists of multiple choice values for output type

[RobinPicard]

When a multiple choice (either through Literal or Enum) is contained in a List as an output type, the value of the multiple choice is missing quotation marks. The problem does not exist with only a multiple choice as an output type as the llm response is a string.

Example to illustrate the problem and to reproduce:

from typing import List, Literal
import outlines
import transformers
from outlines.types.dsl import to_regex, python_types_to_terms
from enum import Enum

TEST_MODEL = "microsoft/Phi-3-mini-4k-instruct"

model = outlines.from_transformers(
    transformers.AutoModelForCausalLM.from_pretrained(TEST_MODEL),
    transformers.AutoTokenizer.from_pretrained(TEST_MODEL),
)

# multiple choices (same issue with Enum as with Literal)
output_type = List[Literal["Paris", "London", "Rome", "Berlin"]]

print(to_regex(python_types_to_terms(output_type))) # \[(Paris|London|Rome|Berlin)(,\ (Paris|London|Rome|Berlin))*\]

result = model("Give me a list of cities.", output_type, max_new_tokens=100)
print(result) # [Paris]

# string
output_type = List[str]

print(to_regex(python_types_to_terms(output_type))) # \[("[^"]*")(,\ ("[^"]*"))*\]

result = model("Give me a list of cities.", output_type, max_new_tokens=100)
print(result) # ["New York", "Los Angeles", "Chicago", "Houston", "Phoenix"]

I think there may be similar issue of strings in complex output types elsewhere.