Improve `factorial(n, k)` performance for `BigInt`

Use in-place operations from Base.GMP.MPZ to reduce allocations.

With this change I get a massive speed-up, for instance I pass from

julia> @benchmark factorial(big"1000", big"500")
BenchmarkTools.Trial: 10000 samples with 1 evaluation.
 Range (min … max):  100.233 μs … 105.019 ms  ┊ GC (min … max):  0.00% … 80.08%
 Time  (median):     116.240 μs               ┊ GC (median):     0.00%
 Time  (mean ± σ):   259.614 μs ±   2.368 ms  ┊ GC (mean ± σ):  20.95% ±  2.33%

  █▇▇▄▂▁                                                        ▂
  ███████▇▇▇█▅▅▃▃▃▃▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▄ █
  100 μs        Histogram: log(frequency) by time        962 μs <

 Memory estimate: 192.52 KiB, allocs estimate: 3002.

julia> @benchmark factorial(big"1000", big"500")
BenchmarkTools.Trial: 10000 samples with 991 evaluations.
 Range (min … max):   49.491 ns … 92.862 μs  ┊ GC (min … max):  0.00% … 52.86%
 Time  (median):      54.742 ns              ┊ GC (median):     0.00%
 Time  (mean ± σ):   110.839 ns ±  2.088 μs  ┊ GC (mean ± σ):  23.97% ±  1.28%

   ▇█▆▁
  ▂█████▆▆▅▃▂▂▃▅▇▆▄▄▃▃▂▂▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ ▂
  49.5 ns         Histogram: frequency by time          105 ns <

 Memory estimate: 40 bytes, allocs estimate: 2.

for an approximate 2000x performance boost.

JuliaMath / Combinatorics.jl

Improve `factorial(n, k)` performance for `BigInt` #149

Codecov Report