x86: zero extend the 32-bit length in int32_to_float_fmul_scalar implicitly

This reverts commit 5dfe4edad6.

libavcodec/x86/fmtconvert.asm

@@ -37,12 +37,9 @@ cglobal int32_to_float_fmul_scalar, 4, 4, %1, dst, src, mul, len
     SWAP 0, 2
 %elif ARCH_X86_32
     movss   m0, mulm
-%endif
-%if ARCH_X86_64
-    movsxd  lenq, lend
 %endif
     SPLATD  m0
-    shl     lenq, 2
+    shl     lend, 2
     add     srcq, lenq
     add     dstq, lenq
     neg     lenq