Don't use mpz_powm()

[skirpichev]

Closes #169

[skirpichev]

No measurable difference with benchmark:

# bench.py
import pyperf
import random
from gmp import mpz

random.seed(1)
lbits = 64
runner = pyperf.Runner()
a, b, c = 123, 11, 7
runner.bench_func(f'pow({a}, {b}, {c})', pow, a, b, c)
c *= 4
runner.bench_func(f'pow({a}, {b}, {c})', pow, a, b, c)
a, b, c = map(mpz, (random.randint(1<<(7*lbits), 1<<(11*lbits))
                    for _ in range(3)))
mbits = max(map(lambda x: x.bit_length(), [a, b, c]))
runner.bench_func(f'pow(a, b, c) with max {mbits}-bits', pow, a, b, c)