feat(dec): Decaps D0 - op_i widen + dk/c load + parse

Scaffolding for ML-KEM Decaps (FIPS 203 Alg 18):
- op_i widened to 2-bit: 00=KeyGen, 01=Encaps, 10=Decaps (op_r too).
- New ST_DEC_LOAD state (D0: settles to DONE so load/parse is dbg-checkable).
- dk (=sk) streamed via dk_in_*; load logic routes each byte by region:
  [0,384K)->dk_pke (dkp_bram), [384K,768K+32)->ek_pke (ek_bram),
  [768K+32,+32)->H(ek) (hek_r), [768K+64,+32)->z (z_r). Routing uses the
  LIVE k_i input, not start-captured k_r (dk is streamed before start_i).
- c (=ct) streamed via c_in_* into a SEPARATE c_in_bram, so the computed c'
  (ct_bram) can later be compared against original c and J(z||c) can read c.
- New dbg taps: dbg_mprime_o/dbg_kbar_o/dbg_decz_o/dbg_dech_o.

TB: tb_mlkem_dec_katK_xsim verifies dk parse (H(ek), z, ek_pke/dk_pke BRAM
round-trip). gen_decaps_vectors.py emits dec_k{K}_c{N}_{dk,ct,ss,ctn,ssn}.hex
from the NIST KAT. run_tb.sh gains a 'dec' module (mirrors 'enc').

Regression fix: old KeyGen/Encaps TBs didn't connect the new input ports,
floating them to X and corrupting the ek/dkp write muxes -> tied off
dk_in_*/c_in_*/new dbg taps in both.

Verified: dec D0 K=2/3/4 PASS; KeyGen K=2 + Encaps K=2 unregressed.

run_tb.sh

@@ -10,12 +10,15 @@
 #        ./run_tb.sh enc                  # ML-KEM Encaps: all K, all cases (0..2)
 #        ./run_tb.sh enc 2                # Encaps K=2, all its cases
 #        ./run_tb.sh enc 4 1              # Encaps K=4, only CASE=1
+#        ./run_tb.sh dec                  # ML-KEM Decaps: all K, all cases (0..2)
+#        ./run_tb.sh dec 2 0              # Decaps K=2, only CASE=0
 #        ./run_tb.sh --list
 #
-# 'top' (KeyGen) and 'enc' (Encaps) share the same RTL datapath; both compile
-# the xvlog lines from sync_rtl/top/TB/xsim_run.tcl. They differ only in the
-# testbench: top -> tb_mlkem_kg_katK_xsim, enc -> tb_mlkem_enc_katK_xsim.
-# For these two, K (2/3/4) and CASE select a single KAT run so you can iterate
+# 'top' (KeyGen), 'enc' (Encaps) and 'dec' (Decaps) share the same RTL datapath;
+# all compile the xvlog lines from sync_rtl/top/TB/xsim_run.tcl. They differ only
+# in the testbench: top -> tb_mlkem_kg_katK_xsim, enc -> tb_mlkem_enc_katK_xsim,
+# dec -> tb_mlkem_dec_katK_xsim.
+# For these, K (2/3/4) and CASE select a single KAT run so you can iterate
 # quickly. Other modules ignore the extra args and run their xsim_run.tcl verbatim.
 #
 # Prerequisites:
@@ -35,6 +38,7 @@ if [ "$1" = "--list" ]; then
         fi
     done
     echo "  enc   (ML-KEM Encaps; shares the 'top' RTL/tcl, enc testbench)"
+    echo "  dec   (ML-KEM Decaps; shares the 'top' RTL/tcl, dec testbench)"
     exit 0
 fi
 
@@ -48,8 +52,8 @@ MODULE="$1"
 SEL_K="$2"          # optional: 2/3/4 (top/enc modules only)
 SEL_CASE="$3"       # optional: KAT case index (top/enc modules only)
 # 'enc' is not its own RTL dir; it reuses the 'top' KeyGen tcl compile list
-# (same datapath) and swaps in the encaps testbench.
-if [ "$MODULE" = "enc" ]; then
+# (same datapath) and swaps in the encaps testbench. 'dec' likewise (Decaps).
+if [ "$MODULE" = "enc" ] || [ "$MODULE" = "dec" ]; then
     TB_DIR="$SCRIPT_DIR/sync_rtl/top/TB"
 else
     TB_DIR="$SCRIPT_DIR/sync_rtl/$MODULE/TB"
@@ -188,11 +192,59 @@ run_enc_selected() {
     return $fail
 }
 
+# ML-KEM Decaps runner. Same compile basis as enc (top tcl minus KeyGen TB) plus
+# the decaps TB. Result line matches the final-stage PASS tag (D0..D7).
+run_dec_selected() {
+    local tcl_file="$1" ksel="$2" csel="$3"
+    set +e
+
+    rm -rf xsim.dir .Xil
+
+    while read -r cmd; do
+        [[ "$cmd" == *tb_mlkem_kg_katK* ]] && continue
+        eval "$cmd" || { echo "COMPILE FAILED: $cmd"; return 1; }
+    done < <(grep -E '^xvlog ' "$tcl_file")
+
+    echo "  xvlog -sv --relax sync_rtl/top/TB/tb_mlkem_dec_katK_xsim.v"
+    xvlog -sv --relax sync_rtl/top/TB/tb_mlkem_dec_katK_xsim.v \
+        || { echo "DEC TB COMPILE FAILED"; return 1; }
+
+    local ks; if [ -n "$ksel" ]; then ks="$ksel"; else ks="2 3 4"; fi
+    local fail=0
+    for k in $ks; do
+        echo "  xelab tb_mlkem_dec_katK_xsim -generic_top KP=$k -s mlkem_dec_k$k --timescale 1ns/1ps"
+        xelab tb_mlkem_dec_katK_xsim -generic_top KP=$k -s mlkem_dec_k$k --timescale 1ns/1ps \
+            || { echo "ELAB FAILED for K=$k"; fail=1; continue; }
+
+        local cases="0 1 2"
+        if [ -n "$csel" ]; then cases="$csel"; fi
+        for c in $cases; do
+            local log="/tmp/run_tb_dec_k${k}_c${c}.log"
+            echo "  xsim mlkem_dec_k$k -R -testplusarg CASE=$c"
+            echo "========================================" | tee "$log"
+            xsim "mlkem_dec_k$k" -R -testplusarg "CASE=$c" 2>&1 | tee -a "$log"
+            echo "========================================" | tee -a "$log"
+            local pf nf
+            # match the highest-stage result line: 'K=.. CASE .. PASS (Dn): ...'
+            pf=$(grep -oE 'PASS \(D[0-7]\)|FAIL \(D[0-7]\)' "$log" | tail -1)
+            nf=$(grep -c 'cannot be opened' "$log")
+            echo "    K=$k CASE=$c -> ${pf:-NORESULT}  (file-not-found=$nf, log: $log)"
+            { [[ "$pf" == PASS* ]] && [ "$nf" -eq 0 ]; } || fail=1
+        done
+    done
+    return $fail
+}
+
 if [ "$MODULE" = "enc" ]; then
     run_enc_selected "$TCL_FILE" "$SEL_K" "$SEL_CASE"
     exit $?
 fi
 
+if [ "$MODULE" = "dec" ]; then
+    run_dec_selected "$TCL_FILE" "$SEL_K" "$SEL_CASE"
+    exit $?
+fi
+
 if [ "$MODULE" = "top" ] && [ -n "$SEL_K" ]; then
     run_top_selected "$TCL_FILE" "$SEL_K" "$SEL_CASE"
     exit $?