Final release

Added documentation

Optimized gf2x_scalarprod when calculating left operands

Shifting results in place for f_sum and g_sum

Now the left operands (f_sum and g_sum) are calculated and shifted directly in their final position, without needing to shift in a temporary location and then copy them 
Only shifting bits, the DIGIT shift once performed by the right_bit_shift_wide_n is now taken care by the offset calculation

Optimized gf2x_scalarprod when recombining the results

Storing values directly in the right locations

If, upon a scalarprod operation, the result was copied as-is to another location, now we skip this step, storing the result in the right location (e.g. in P_xx instead of passing by recombine)

Autogenerated jumpdivstep procedures

The body of the jumpdivstep function has been automatically generated, removing all the control structures and producing a sequence of basic operations on the operands

Iterative version

The original recursive code has been transformed into an iterative version by reconstructing the call stack artificially

Original code

v00

Changed -O2 into -O3