Reduce access latency of GPIO blocks

Rework the `gpio_core` to provide response data in the same cycle
as the request to reduce the access latency.

Reduce the value of the `AccessLatency` parameter of the TL-UL adapter
used to connect the GPIO blocks to the system bus,
to match the `gpio_core` read latency changes.

Author: elliotb-lowrisc

rtl/ip/gpio/rtl/gpio.sv

@@ -67,7 +67,7 @@ module gpio #(
   );
 
   tlul_adapter_reg #(
-    .AccessLatency ( 1            ),
+    .AccessLatency ( 0            ),
     .RegAw         ( RegAddrWidth )
   ) gpio_device_adapter (
     .clk_i,

rtl/ip/gpio/rtl/gpio_core.sv

@@ -115,13 +115,11 @@ module gpio_core #(
     if (!rst_ni) begin
       gp_o              <= '0;
       gp_o_en           <= '0;
-      device_rvalid_o   <= '0;
       pcint_mask        <= '0;
       pcint_enable      <= '0;
       pcint_mode        <= '0;
       pcint_i_sel       <= '0;
       pcint_status      <= '0;
-      rd_reg_idx        <= '0;
     end else begin
       gp_o         <= (gp_o_sel && device_we_i)       ? gp_o_d         : gp_o;
       gp_o_en      <= (gp_o_en_sel && device_we_i)    ? gp_o_en_d      : gp_o_en;
@@ -138,12 +136,12 @@ module gpio_core #(
       end else begin
         pcint_status <= pcint_status;
       end
-
-      device_rvalid_o <= device_req_i && !device_we_i;
-      rd_reg_idx <= reg_addr[RegAddr-1:2];
     end
   end
 
+  assign device_rvalid_o = device_req_i && !device_we_i;
+  assign rd_reg_idx = reg_addr[RegAddr-1:2];
+
   logic [3:0] unused_o_device_be;
   logic [3:0] unused_i_device_be;