Incorrect float to fx conversion in decoder app with object edit file interface

Basic info

• Float reference:
  • Decoder (float):0904f1f2
• Fixed point:
  • Encoder (fixed):43b7b28d
  • Decoder (fixed):43b7b28d

Bug description

When reading object editing input from file, the float input is casted to Word32 before converting float values to correct integer range. I.e., float input is first rounded to nearest integer (1.2f -> 1) and after that shifted to correct fx range, causing incorrect input.

Ways to reproduce

* ENC *
-ism_masa
4
2
../scripts/test/stvISM1.csv
../scripts/test/stvISM2.csv
../scripts/test/stvISM3.csv
../scripts/test/stvISM4.csv
../scripts/test/stv2MASA2TC48c.met
96000
48
../scripts/test/stvOMASA_4ISM_2MASA2TC48c.wav
bit


* DEC * 
./IVAS_dec -obj_edit
../scripts/object_edit/combined_edit.txt
BINAURAL
48
bit
out.wav