Default add_generation_prompt for Chat Templates to true in Python API (microsoft#1539)

Adds default to make `add_generation_prompt` optional in the Python API
(note: because of lack of support for optionals in the C API we can only
do this for Python).

---------

Co-authored-by: Sayan Shaw <[email protected]>
Co-authored-by: Baiju Meswani <[email protected]>

Author: 3 people

src/python/python.cpp

@@ -358,7 +358,7 @@ PYBIND11_MODULE(onnxruntime_genai, m) {
       })
       .def("to_token_id", &OgaTokenizer::ToTokenId)
       .def("decode", [](const OgaTokenizer& t, pybind11::array_t<int32_t> tokens) -> std::string { return t.Decode(ToSpan(tokens)).p_; })
-      .def("apply_chat_template", [](const OgaTokenizer& t, const char* template_str, const char* messages, const char* tools, bool add_generation_prompt) -> std::string { return t.ApplyChatTemplate(template_str, messages, tools, add_generation_prompt).p_; }, pybind11::arg("template_str") = nullptr, pybind11::arg("messages"), pybind11::arg("tools") = nullptr, pybind11::arg("add_generation_prompt"))
+      .def("apply_chat_template", [](const OgaTokenizer& t, const char* messages, const char* template_str, const char* tools, bool add_generation_prompt) -> std::string { return t.ApplyChatTemplate(template_str, messages, tools, add_generation_prompt).p_; }, pybind11::arg("messages"), pybind11::kw_only(), pybind11::arg("template_str") = nullptr, pybind11::arg("tools") = nullptr, pybind11::arg("add_generation_prompt") = true)
       .def("encode_batch", [](const OgaTokenizer& t, std::vector<std::string> strings) {
         std::vector<const char*> c_strings;
         for (const auto& s : strings)