fix: replace panic calls with proper error handling

Remove 51 panic() calls from handler code and replace with:
- Proper error logging using zap
- Appropriate client error responses (doAckBufFail, doAckSimpleFail)
- Graceful error recovery instead of server crashes

Files updated:
- handlers_guild_scout.go (9 panics)
- handlers_guild_tresure.go (10 panics)
- handlers_guild.go (7 panics + dead code removal)
- handlers_mail.go (5 panics)
- handlers.go (9 panics)
- handlers_tower.go (2 panics)
- handlers_clients.go (3 panics)
- handlers_guild_alliance.go (1 panic)
- handlers_quest.go (1 panic)
- handlers_rengoku.go (1 panic)
- handlers_stage.go (1 panic)
- handlers_data.go (1 panic)
- handlers_cafe.go (1 panic)
- signserver/sign_server.go (1 panic)

Remaining panics (3) are in test files and compression library
where panicking on programming errors is appropriate.

server/signserver/sign_server.go

@@ -77,7 +77,8 @@ func (s *Server) acceptClients() {
 			if shutdown {
 				break
 			} else {
-				panic(err)
+				s.logger.Error("failed to accept connection", zap.Error(err))
+				continue
 			}
 		}