Pull action_log into its own crate (#35959)

Release Notes:

- N/A

crates/agent/src/agent_profile.rs

@@ -326,7 +326,7 @@ mod tests {
             _input: serde_json::Value,
             _request: Arc<language_model::LanguageModelRequest>,
             _project: Entity<Project>,
-            _action_log: Entity<assistant_tool::ActionLog>,
+            _action_log: Entity<action_log::ActionLog>,
             _model: Arc<dyn language_model::LanguageModel>,
             _window: Option<gpui::AnyWindowHandle>,
             _cx: &mut App,

crates/agent/src/context_server_tool.rs

@@ -1,7 +1,8 @@
 use std::sync::Arc;
 
+use action_log::ActionLog;
 use anyhow::{Result, anyhow, bail};
-use assistant_tool::{ActionLog, Tool, ToolResult, ToolSource};
+use assistant_tool::{Tool, ToolResult, ToolSource};
 use context_server::{ContextServerId, types};
 use gpui::{AnyWindowHandle, App, Entity, Task};
 use icons::IconName;

crates/agent/src/thread.rs

@@ -8,9 +8,10 @@ use crate::{
     },
     tool_use::{PendingToolUse, ToolUse, ToolUseMetadata, ToolUseState},
 };
+use action_log::ActionLog;
 use agent_settings::{AgentProfileId, AgentSettings, CompletionMode, SUMMARIZE_THREAD_PROMPT};
 use anyhow::{Result, anyhow};
-use assistant_tool::{ActionLog, AnyToolCard, Tool, ToolWorkingSet};
+use assistant_tool::{AnyToolCard, Tool, ToolWorkingSet};
 use chrono::{DateTime, Utc};
 use client::{ModelRequestUsage, RequestUsage};
 use cloud_llm_client::{CompletionIntent, CompletionRequestStatus, Plan, UsageLimit};