Add support for integers and tool configuration in schema handling

This update introduces support for `jsonschema.Integer` types and updates the logic to handle nested items in schemas. Added a new default error log for unknown types using `slog.Error`. Also, integrated tool configuration with a `FunctionCallingConfig` when `dontRequireTool` is false.

request.go

@@ -1,6 +1,8 @@
 package go_llm
 
-import "github.com/sashabaranov/go-openai"
+import (
+	"github.com/openai/openai-go"
+)
 
 type rawAble interface {
 	toRaw() map[string]any
@@ -8,13 +10,13 @@ type rawAble interface {
 }
 
 type Input interface {
-	toChatCompletionMessages() []openai.ChatCompletionMessage
+	toChatCompletionMessages(model string) []openai.ChatCompletionMessageParamUnion
 }
 type Request struct {
 	Conversation []Input
 	Messages     []Message
 	Toolbox      *ToolBox
-	Temperature  *float32
+	Temperature  *float64
 }
 
 // NextRequest will take the current request's conversation, messages, the response, and any tool results, and
@@ -33,16 +35,8 @@ func (req Request) NextRequest(resp ResponseChoice, toolResults []ToolCallRespon
 		res.Conversation = append(res.Conversation, msg)
 	}
 
-	// if there are tool calls, then we need to add those to the conversation
-	for _, call := range resp.Calls {
-		res.Conversation = append(res.Conversation, call)
-	}
-
-	if resp.Content != "" || resp.Refusal != "" {
-		res.Conversation = append(res.Conversation, Message{
-			Role: RoleAssistant,
-			Text: resp.Content,
-		})
+	if resp.Content != "" || resp.Refusal != "" || len(resp.Calls) > 0 {
+		res.Conversation = append(res.Conversation, resp)
 	}
 
 	// if there are tool results, then we need to add those to the conversation