Fix Set_Input_Pos feed-forward field widths

[LaPhilosophie]

Set_Input_Pos defines both Vel_FF and Torque_FF as signed 16-bit fields in the CAN protocol, but the position-control callback writes each as int8_t. This leaves the high byte zero: for example, a velocity feed-forward of -0.125 rev/s produces 83 00 (+131 counts) instead of 83 FF (-125 counts).

Use int16_t for both conversions and writes, matching the protocol and the existing Set_Input_Pos_msg_t encoder. The change is limited to these two lines.

Validation: compiled the original and patched callback with the repository's byte_swap.hpp and a recording CAN transport under GCC 13.3, with undefined-behavior and float-cast-overflow sanitizers. A 255-input sweep produced 127 mismatches before the fix and none afterward. Fourteen additional patched cases varied the fields independently, including values beyond the 8-bit range; both fields decoded correctly and the position field remained unchanged. This was a callback-level test, without a full ROS launch or physical hardware.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[CLAassistant]

All committers have signed the CLA.

[samuelsadok]

Verified that the docs (int16) are correct, and thus this change is also correct, thanks!