[mc] initial pass at multithreaded

[sjkelly]

Partially addresses #24. Performance improvements are across the board good. The single threaded case suffers a bit though.

Performance (median time on Torus)
Master:
SDF - 2.721 ms
Func - 6.745 ms

Threads = 6:
SDF - 2.212 ms
Func - 2.699 ms

Threads = 1:
SDF - 3.818 ms
Func - 7.313 ms

[sjkelly]

Test pass on travis since it is running single thread, but fail locally on multiple threads. I may need to rework the tests to handle the non-determinism of threads.

[sjkelly]

steve@sjkdsk1:~/.julia/dev/Meshing (sjk/threads3)$ export JULIA_NUM_THREADS=6;

steve@sjkdsk1:~/.julia/dev/Meshing (sjk/threads3)$ juliamaster ./bench/benchmark.jl 
Benchmarking Meshing.jl...

SDF Mesh
3-element BenchmarkTools.BenchmarkGroup:
  tags: []
  "MarchingCubes{Float64}" => BenchmarkTools.Trial: 
	  memory estimate:  1.06 MiB
	  allocs estimate:  88
	  --------------
	  minimum time:     1.508 ms (0.00% GC)
	  median time:      1.779 ms (0.00% GC)
	  mean time:        1.804 ms (0.75% GC)
	  maximum time:     3.724 ms (0.00% GC)
	  --------------
	  samples:          2767
	  evals/sample:     1

Function Mesh
2-element BenchmarkTools.BenchmarkGroup:
  tags: []
  "MarchingCubes{Float64}" => BenchmarkTools.Trial: 
	  memory estimate:  1.07 MiB
	  allocs estimate:  171
	  --------------
	  minimum time:     1.917 ms (0.00% GC)
	  median time:      2.211 ms (0.00% GC)
	  mean time:        2.280 ms (0.64% GC)
	  maximum time:     4.649 ms (0.00% GC)
	  --------------
	  samples:          2191
	  evals/sample:     1


steve@sjkdsk1:~/.julia/dev/Meshing (sjk/threads3)$ export JULIA_NUM_THREADS=1;

steve@sjkdsk1:~/.julia/dev/Meshing (sjk/threads3)$ juliamaster ./bench/benchmark.jl 
Benchmarking Meshing.jl...

SDF Mesh
3-element BenchmarkTools.BenchmarkGroup:
  tags: []
  "MarchingCubes{Float64}" => BenchmarkTools.Trial: 
	  memory estimate:  1.05 MiB
	  allocs estimate:  51
	  --------------
	  minimum time:     3.144 ms (0.00% GC)
	  median time:      3.411 ms (0.00% GC)
	  mean time:        4.685 ms (0.91% GC)
	  maximum time:     1.360 s (1.73% GC)
	  --------------
	  samples:          1060
	  evals/sample:     1

Function Mesh
2-element BenchmarkTools.BenchmarkGroup:
  tags: []
  "MarchingCubes{Float64}" => BenchmarkTools.Trial: 
	  memory estimate:  1.07 MiB
	  allocs estimate:  136
	  --------------
	  minimum time:     6.509 ms (0.00% GC)
	  median time:      7.035 ms (0.00% GC)
	  mean time:        7.459 ms (0.27% GC)
	  maximum time:     286.279 ms (0.00% GC)
	  --------------
	  samples:          671
	  evals/sample:     1

steve@sjkdsk1:~/.julia/dev/Meshing (sjk/threads3)$ git checkout master

steve@sjkdsk1:~/.julia/dev/Meshing (master=)$ juliamaster ./bench/benchmark.jl 
Benchmarking Meshing.jl...

SDF Mesh
3-element BenchmarkTools.BenchmarkGroup:
  tags: []
  "MarchingCubes{Float64}" => BenchmarkTools.Trial: 
	  memory estimate:  1.05 MiB
	  allocs estimate:  40
	  --------------
	  minimum time:     2.786 ms (0.00% GC)
	  median time:      3.097 ms (0.00% GC)
	  mean time:        4.336 ms (0.92% GC)
	  maximum time:     1.390 s (1.43% GC)
	  --------------
	  samples:          1144
	  evals/sample:     1
Function Mesh
2-element BenchmarkTools.BenchmarkGroup:
  tags: []
  "MarchingCubes{Float64}" => BenchmarkTools.Trial: 
	  memory estimate:  1.05 MiB
	  allocs estimate:  46
	  --------------
	  minimum time:     5.757 ms (0.00% GC)
	  median time:      6.217 ms (0.00% GC)
	  mean time:        6.720 ms (0.29% GC)
	  maximum time:     296.505 ms (0.00% GC)
	  --------------
	  samples:          744
	  evals/sample:     1