Fix: wait for ALL 4 multipliers to finish before accumulation

rtl/ip/matmul/matmul_axi_slave.v

@@ -428,8 +428,7 @@ module matmul_axi_slave (
                 
                 C_WAIT_MULT: begin
                     mult_start <= 1'b0;
-                    // 4个乘法器同时启动，监控第一个即可
+                    if (mult_done[0] && mult_done[1] && mult_done[2] && mult_done[3]) begin
-                    if (mult_done[0]) begin
                         state <= C_ACCUM;
                     end
                 end