feat: add relevance-gated proactive replies

Replace random-only proactive reply logic with LLM relevance check.
The bot now evaluates recent conversation context and user memory
before deciding to jump in, then applies reply_chance as a second
gate. Bump reply_chance values higher since the relevance filter
prevents most irrelevant replies.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

config.yaml

@@ -83,7 +83,7 @@ modes:
     description: "Friendly chat participant"
     prompt_file: "personalities/chat_chatty.txt"
     proactive_replies: true
-    reply_chance: 0.10
+    reply_chance: 0.40
     moderation: relaxed
     relaxed_thresholds:
       warning_threshold: 0.80
@@ -96,7 +96,7 @@ modes:
     description: "Savage roast mode"
     prompt_file: "personalities/chat_roast.txt"
     proactive_replies: true
-    reply_chance: 0.20
+    reply_chance: 0.60
     moderation: relaxed
     relaxed_thresholds:
       warning_threshold: 0.85
@@ -109,7 +109,7 @@ modes:
     description: "Your biggest fan"
     prompt_file: "personalities/chat_hype.txt"
     proactive_replies: true
-    reply_chance: 0.15
+    reply_chance: 0.50
     moderation: relaxed
     relaxed_thresholds:
       warning_threshold: 0.80
@@ -122,7 +122,7 @@ modes:
     description: "Had a few too many"
     prompt_file: "personalities/chat_drunk.txt"
     proactive_replies: true
-    reply_chance: 0.20
+    reply_chance: 0.60
     moderation: relaxed
     relaxed_thresholds:
       warning_threshold: 0.85
@@ -135,7 +135,7 @@ modes:
     description: "Insufferable grammar nerd mode"
     prompt_file: "personalities/chat_english_teacher.txt"
     proactive_replies: true
-    reply_chance: 0.20
+    reply_chance: 0.60
     moderation: relaxed
     relaxed_thresholds:
       warning_threshold: 0.85