Have read_file support images (#30435)

This is very basic support for them. There are a number of other TODOs
before this is really a first-class supported feature, so not adding any
release notes for it; for now, this PR just makes it so that if
read_file tries to read a PNG (which has come up in practice), it at
least correctly sends it to Anthropic instead of messing up.

This also lays the groundwork for future PRs for more first-class
support for images in tool calls across more image file formats and LLM
providers.

Release Notes:

- N/A

---------

Co-authored-by: Agus Zubiaga <hi@aguz.me>
Co-authored-by: Agus Zubiaga <agus@zed.dev>

crates/language_models/src/provider/open_ai.rs

@@ -12,7 +12,8 @@ use language_model::{
     AuthenticateError, LanguageModel, LanguageModelCompletionError, LanguageModelCompletionEvent,
     LanguageModelId, LanguageModelName, LanguageModelProvider, LanguageModelProviderId,
     LanguageModelProviderName, LanguageModelProviderState, LanguageModelRequest,
-    LanguageModelToolChoice, LanguageModelToolUse, MessageContent, RateLimiter, Role, StopReason,
+    LanguageModelToolChoice, LanguageModelToolResultContent, LanguageModelToolUse, MessageContent,
+    RateLimiter, Role, StopReason,
 };
 use open_ai::{Model, ResponseStreamEvent, stream_completion};
 use schemars::JsonSchema;
@@ -295,6 +296,10 @@ impl LanguageModel for OpenAiLanguageModel {
         true
     }
 
+    fn supports_images(&self) -> bool {
+        false
+    }
+
     fn supports_tool_choice(&self, choice: LanguageModelToolChoice) -> bool {
         match choice {
             LanguageModelToolChoice::Auto => true,
@@ -392,8 +397,16 @@ pub fn into_open_ai(
                     }
                 }
                 MessageContent::ToolResult(tool_result) => {
+                    let content = match &tool_result.content {
+                        LanguageModelToolResultContent::Text(text) => text.to_string(),
+                        LanguageModelToolResultContent::Image(_) => {
+                            // TODO: Open AI image support
+                            "[Tool responded with an image, but Zed doesn't support these in Open AI models yet]".to_string()
+                        }
+                    };
+
                     messages.push(open_ai::RequestMessage::Tool {
-                        content: tool_result.content.to_string(),
+                        content,
                         tool_call_id: tool_result.tool_use_id.to_string(),
                     });
                 }