AArch64: Integrate scaling into first invNTT layers

Previously, the initial scaling step in the inverseNTT
was handled separately and without SLOTHY optimization.

This commit integrates scaling into the initial layer
of the invNTT and re-runs SLOTHY.

Signed-off-by: Hanno Becker <beckphan@amazon.co.uk>

Author: hanno-becker

dev/aarch64_clean/src/intt.S

@@ -235,7 +235,8 @@ MLK_ASM_FN_SYMBOL(intt_asm)
         mov inp, in
         mov count, #8
 
-intt_scale_start:
+        .p2align 2
+intt_layer4567_start:
 
         ldr q_data0, [inp, #(16*0)]
         ldr q_data1, [inp, #(16*1)]
@@ -245,25 +246,6 @@ intt_scale_start:
         mul_ninv data0, data1, data2, data3, data0, data1, data2, data3
         // Bounds: Absolute value < q
 
-        str q_data0, [inp], #64
-        str q_data1, [inp, #(-64 + 16*1)]
-        str q_data2, [inp, #(-64 + 16*2)]
-        str q_data3, [inp, #(-64 + 16*3)]
-
-        subs count, count, #1
-        cbnz count, intt_scale_start
-
-        mov inp, in
-        mov count, #8
-
-        .p2align 2
-intt_layer4567_start:
-
-        ldr q_data0, [inp, #(16*0)]
-        ldr q_data1, [inp, #(16*1)]
-        ldr q_data2, [inp, #(16*2)]
-        ldr q_data3, [inp, #(16*3)]
-
         transpose4 data // manual ld4
 
         load_next_roots_67