🚧 Thank you for opening a PR!

• We should be able to understand what the PR does from its title only;
• There is a high-level description of the changes;
• If I add a new feature, there is an issue discussing it already; — bug fix, no feature
• There are links to all the relevant issues, discussions and PRs;
• The branch is rebased on the latest main commit;
• Commit messages follow these guidelines;
• One commit per logical change;
• The code respects the current naming conventions;
• Docstrings follow the numpy style guide;
• pre-commit is installed and configured on your machine, and you ran it before opening the PR;
• There are tests covering the changes;
• The documentation is up-to-date; — no documented behaviour changes

Summary

_ensure_json_quoted now encodes String values with json.dumps instead of
wrapping them in quote characters by hand, so a literal containing ", \ or
a control character produces the escape sequence JSON requires.

Why

_ensure_json_quoted (src/outlines/types/dsl.py:880) built the quoted form as
String(f'"{term.value}"'). That adds the delimiters but never escapes the
value, so the generated regex only matched the unescaped text. For
list[Literal['say "hi"']] the pattern came out as:

\[("say\ "hi"")(,\ ("say\ "hi""))*\]

re.fullmatch on ["say \"hi\""] — the encoding json.dumps produces, and the
only form a JSON parser accepts — returns None. The pattern instead accepts
["say "hi""], which no parser can read. dict[Literal['say "hi"'], str] is
wrong the same way for its keys, and Tuple[Literal[...]] for its elements,
since all three go through the same helper.

A literal containing a backslash, a tab or a newline is in the same position:
_json_string_builtin_patterns in #2015 and the corresponding section of #1962
already note that these values need escaping on top of quoting; that PR scopes
itself to built-in Regex terms and leaves the String branch untouched, so
this is the other half.

Changes

• src/outlines/types/dsl.py: _ensure_json_quoted returns
  String(json.dumps(term.value, ensure_ascii=False)).
• tests/types/test_dsl.py: two tests covering List, Dict and Tuple.

ensure_ascii=False keeps non-ASCII characters literal, matching the previous
output and the outlines_core JSON-Schema path; json.dumps with ASCII values
reproduces the old string exactly, so nothing outside the " / \ /
control-character cases changes.

Testing

$ .venv/bin/python -m pytest tests/types/test_dsl.py -k "json_escapes"   # before the fix
FAILED tests/types/test_dsl.py::test_e2e_list_literal_requires_json_escapes
FAILED tests/types/test_dsl.py::test_e2e_dict_and_tuple_literal_require_json_escapes
2 failed, 60 deselected

$ .venv/bin/python -m pytest tests/types/test_dsl.py -k "json_escapes"   # after
2 passed, 60 deselected

$ .venv/bin/python -m pytest tests/types/
348 passed in 0.93s

$ .venv/bin/pre-commit run --all-files
check for merge conflicts...................Passed
debug statements (python)...................Passed
fix end of files............................Passed
trim trailing whitespace....................Passed
mypy........................................Passed
ruff........................................Passed

No issue is filed for this; it is the String half of the escaping gap
described in #1962, tracked there and explicitly left open by #2015.