re-use Encoder::op_num

[enricozb]

The Instruction::OpNum optimization was accidentally removed in #84.

This PR re-introduces the optimization.

[HigherOrderBot]

Perf run for 9caffc6:

file              mode        main          9caffc6dc1d7
========================================================
merge_sort        intr-singl       7.258 s       7.186 s
                  intr-multi       3.564 s       3.742 s
--------------------------------------------------------
boom              intr-singl       2.227 s       2.282 s
                  intr-multi       2.348 s       1.602 s
--------------------------------------------------------
bitonic_sort_lam  intr-singl      11.089 s      11.553 s
                  intr-multi       5.520 s       5.607 s
--------------------------------------------------------
radix_sort_lam    intr-singl       9.311 s       9.604 s
                  intr-multi       4.628 s       4.630 s
--------------------------------------------------------
sum_tree          intr-singl       9.132 s       9.089 s
                  intr-multi       4.364 s       4.433 s
--------------------------------------------------------
c2                intr-singl       0.000 s       0.000 s
                  intr-multi       0.001 s       0.001 s
--------------------------------------------------------
sum_tail          intr-singl       1.119 s       1.133 s
                  intr-multi       1.130 s       1.119 s
--------------------------------------------------------
sum_rec           intr-singl      14.332 s      13.451 s
                  intr-multi       7.056 s       6.960 s