Implement AI memory system with psychological priority scoring

Core changes:
- Add AI interpreter module for content interpretation and priority scoring
- Extend Memory struct with interpreted_content, priority_score (f32: 0.0-1.0), and user_context
- Implement automatic memory pruning based on priority scores
- Add capacity management (default: 100 memories max)
- Create comprehensive design documentation

Technical details:
- Changed priority_score from u8 (1-100) to f32 (0.0-1.0) for better AI compatibility
- Add create_memory_with_ai() method for AI-enhanced memory creation
- Implement get_memories_by_priority() for priority-based sorting
- Score evaluation criteria: emotional impact, user relevance, novelty, utility

Philosophy:
This implements a "psychological priority memory system" where AI interprets
and evaluates memories rather than storing raw content. Inspired by how human
memory works - interpreting and prioritizing rather than perfect recording.

src/ai_interpreter.rs

@@ -0,0 +1,143 @@
+use anyhow::{Context, Result};
+
+#[cfg(feature = "ai-analysis")]
+use openai::{
+    chat::{ChatCompletion, ChatCompletionMessage, ChatCompletionMessageRole},
+    set_key,
+};
+
+pub struct AIInterpreter {
+    #[cfg(feature = "ai-analysis")]
+    api_key: Option<String>,
+}
+
+impl AIInterpreter {
+    pub fn new() -> Self {
+        #[cfg(feature = "ai-analysis")]
+        {
+            let api_key = std::env::var("OPENAI_API_KEY").ok();
+            if let Some(ref key) = api_key {
+                set_key(key.clone());
+            }
+            AIInterpreter { api_key }
+        }
+        #[cfg(not(feature = "ai-analysis"))]
+        {
+            AIInterpreter {}
+        }
+    }
+
+    /// AI解釈: 元のコンテンツを解釈して新しい表現を生成
+    #[cfg(feature = "ai-analysis")]
+    pub async fn interpret_content(&self, content: &str) -> Result<String> {
+        if self.api_key.is_none() {
+            return Ok(content.to_string());
+        }
+
+        let messages = vec![
+            ChatCompletionMessage {
+                role: ChatCompletionMessageRole::System,
+                content: Some("あなたは記憶を解釈するAIです。与えられたテキストを解釈し、より深い意味や文脈を抽出してください。元のテキストの本質を保ちながら、新しい視点や洞察を加えてください。".to_string()),
+                name: None,
+                function_call: None,
+            },
+            ChatCompletionMessage {
+                role: ChatCompletionMessageRole::User,
+                content: Some(format!("以下のテキストを解釈してください:\n\n{}", content)),
+                name: None,
+                function_call: None,
+            },
+        ];
+
+        let chat_completion = ChatCompletion::builder("gpt-3.5-turbo", messages.clone())
+            .create()
+            .await
+            .context("Failed to create chat completion")?;
+
+        let response = chat_completion
+            .choices
+            .first()
+            .and_then(|choice| choice.message.content.clone())
+            .unwrap_or_else(|| content.to_string());
+
+        Ok(response)
+    }
+
+    #[cfg(not(feature = "ai-analysis"))]
+    pub async fn interpret_content(&self, content: &str) -> Result<String> {
+        Ok(content.to_string())
+    }
+
+    /// 心理判定: テキストの重要度を0.0-1.0のスコアで評価
+    #[cfg(feature = "ai-analysis")]
+    pub async fn calculate_priority_score(&self, content: &str, user_context: Option<&str>) -> Result<f32> {
+        if self.api_key.is_none() {
+            return Ok(0.5); // デフォルトスコア
+        }
+
+        let context_info = user_context
+            .map(|ctx| format!("\n\nユーザーコンテキスト: {}", ctx))
+            .unwrap_or_default();
+
+        let messages = vec![
+            ChatCompletionMessage {
+                role: ChatCompletionMessageRole::System,
+                content: Some(format!(
+                    "あなたは記憶の重要度を評価するAIです。以下の基準で0.0-1.0のスコアをつけてください:\n\
+                    - 感情的インパクト (0.0-0.25)\n\
+                    - ユーザーとの関連性 (0.0-0.25)\n\
+                    - 新規性・独自性 (0.0-0.25)\n\
+                    - 実用性 (0.0-0.25)\n\n\
+                    スコアのみを小数で返してください。例: 0.75{}", context_info
+                )),
+                name: None,
+                function_call: None,
+            },
+            ChatCompletionMessage {
+                role: ChatCompletionMessageRole::User,
+                content: Some(format!("以下のテキストの重要度を評価してください:\n\n{}", content)),
+                name: None,
+                function_call: None,
+            },
+        ];
+
+        let chat_completion = ChatCompletion::builder("gpt-3.5-turbo", messages.clone())
+            .create()
+            .await
+            .context("Failed to create chat completion")?;
+
+        let response = chat_completion
+            .choices
+            .first()
+            .and_then(|choice| choice.message.content.clone())
+            .unwrap_or_else(|| "0.5".to_string());
+
+        // スコアを抽出（小数を含む数字）
+        let score = response
+            .trim()
+            .parse::<f32>()
+            .unwrap_or(0.5)
+            .min(1.0)
+            .max(0.0);
+
+        Ok(score)
+    }
+
+    #[cfg(not(feature = "ai-analysis"))]
+    pub async fn calculate_priority_score(&self, _content: &str, _user_context: Option<&str>) -> Result<f32> {
+        Ok(0.5) // デフォルトスコア
+    }
+
+    /// AI解釈と心理判定を同時に実行
+    pub async fn analyze(&self, content: &str, user_context: Option<&str>) -> Result<(String, f32)> {
+        let interpreted = self.interpret_content(content).await?;
+        let score = self.calculate_priority_score(content, user_context).await?;
+        Ok((interpreted, score))
+    }
+}
+
+impl Default for AIInterpreter {
+    fn default() -> Self {
+        Self::new()
+    }
+}