Expose invocation sources to extension tools (#40382)

## What changed

- Add `ToolCallSource` to extension `ToolCall`s and propagate direct or Code Mode
  invocation metadata, including the runtime cell and nested tool-call IDs.
- Add `ToolCall::response_byte_budget` so extension tools can bound direct-call
  responses by the host truncation policy while allowing Code Mode responses up
  to the tool's own limit.
- Share the skills extension's response-size limit between its `list` and `read`
  tools.

## Testing

- Verify that Code Mode source metadata is preserved when core constructs an
  extension tool call.

GitOrigin-RevId: b10aef01be30d35b67bee3af24a1a4175c9ee112

codex-rs/ext/skills/tests/skills_extension.rs

@@ -24,6 +24,7 @@ use codex_extension_api::SkillInvocationInput;
 use codex_extension_api::SkillInvocationKind;
 use codex_extension_api::ThreadStartInput;
 use codex_extension_api::ToolCall;
+use codex_extension_api::ToolCallSource;
 use codex_extension_api::ToolPayload;
 use codex_extension_api::TurnInputContext;
 use codex_extension_api::WorldStateContributionInput;
@@ -1604,6 +1605,7 @@ async fn skills_list_only_returns_model_visible_bounded_metadata() -> TestResult
             model: "gpt-test".to_string(),
             codex_turn_metadata: None,
             truncation_policy: TruncationPolicy::Bytes(1_024),
+            source: ToolCallSource::Direct,
             conversation_history: ConversationHistory::default(),
             turn_item_emitter: Arc::new(NoopTurnItemEmitter),
             environments: Vec::new(),