Improvements to SP code

Fixed left shifts to be on unsigned types.

Mod exp change to correctly get the highest indeces of exponent -
corrected in some places and now the same in all.

Author: SparkiDev

wolfcrypt/src/sp_arm32.c

@@ -57932,25 +57932,25 @@ L_256_mod_inv_avx2_4_store_done:
         and	edi, 67108863
         sar	edx, 26
         add	esi, edx
-        movsxd	r9, r9d
-        movsxd	r11, r11d
-        movsxd	r13, r13d
-        movsxd	r15, r15d
-        movsxd	rsi, esi
+        movsx	r9, r9d
+        movsx	r11, r11d
+        movsx	r13, r13d
+        movsx	r15, r15d
+        movsx	rsi, esi
         shl	r9, 26
         shl	r11, 26
         shl	r13, 26
         shl	r15, 26
         shl	rsi, 26
-        movsxd	rax, eax
+        movsx	rax, eax
         add	rax, r9
-        movsxd	r10, r10d
+        movsx	r10, r10d
         adc	r10, r11
-        movsxd	r12, r12d
+        movsx	r12, r12d
         adc	r12, r13
-        movsxd	r14, r14d
+        movsx	r14, r14d
         adc	r14, r15
-        movsxd	rdi, edi
+        movsx	rdi, edi
         adc	rdi, rsi
         jge	L_256_mod_inv_avx2_4_no_add_order
         mov	r9, 2756213597218129