Add a basic write benchmark

[AlgorithmsAreCool]

This PR closes #114

I've added a working Benchmark.Net project and some very basic write performance tests.

On my machine they take several minutes to run and produced the following

BenchmarkDotNet=v0.12.1, OS=Windows 10.0.18363.836 (1909/November2018Update/19H2)
Intel Core i7-8750H CPU 2.20GHz (Coffee Lake), 1 CPU, 12 logical and 6 physical cores
.NET Core SDK=5.0.100-preview.5.20279.10
  [Host]     : .NET Core 3.1.5 (CoreCLR 4.700.20.26901, CoreFX 4.700.20.27001), X64 RyuJIT
  DefaultJob : .NET Core 3.1.5 (CoreCLR 4.700.20.26901, CoreFX 4.700.20.27001), X64 RyuJIT

Method	BatchSize	WriteSize	SequentialKeys	Mean	Error	StdDev	Gen 0	Gen 1	Gen 2	Allocated
Write	1	4	True	1.628 μs	0.0220 μs	0.0206 μs	0.0629	-	-	304 B
Write	1	8	True	1.605 μs	0.0101 μs	0.0079 μs	0.0629	-	-	304 B
Write	1	64	True	1.602 μs	0.0075 μs	0.0058 μs	0.0629	-	-	304 B
Write	1	256	True	1.620 μs	0.0210 μs	0.0196 μs	0.0629	-	-	304 B
Write	10	4	True	2.590 μs	0.0098 μs	0.0082 μs	0.0610	-	-	304 B
Write	10	8	True	2.620 μs	0.0333 μs	0.0311 μs	0.0610	-	-	304 B
Write	10	64	True	2.605 μs	0.0264 μs	0.0220 μs	0.0610	-	-	304 B
Write	10	256	True	2.703 μs	0.0282 μs	0.0264 μs	0.0610	-	-	304 B
Write	100	4	True	14.852 μs	0.2910 μs	0.5322 μs	0.0610	-	-	304 B
Write	100	8	True	14.730 μs	0.2944 μs	0.7386 μs	0.0610	-	-	304 B
Write	100	64	True	15.623 μs	0.3049 μs	0.4373 μs	0.0610	-	-	304 B
Write	100	256	True	21.123 μs	0.4208 μs	0.5471 μs	0.0610	-	-	304 B

[CoreyKaylor]

This looks good, I can pull in and merge into a pre-Span branch if you'd like. Obviously we can add more over time but this is a good start.

[AlgorithmsAreCool]

I'll leave that to your discretion, but this at least helps give a baseline to make perf changes against.

[CoreyKaylor]

From my mac on the pre-span branch.

Method	BatchSize	WriteSize	SequentialKeys	Mean	Error	StdDev	Median	Gen 0	Gen 1	Gen 2	Allocated
Write	1	4	True	1.745 μs	0.0194 μs	0.0182 μs	1.738 μs	0.0458	-	-	384 B
Write	1	8	True	1.763 μs	0.0132 μs	0.0110 μs	1.761 μs	0.0458	-	-	384 B
Write	1	64	True	1.843 μs	0.0249 μs	0.0221 μs	1.849 μs	0.0458	-	-	384 B
Write	1	256	True	1.888 μs	0.0376 μs	0.0573 μs	1.910 μs	0.0458	-	-	384 B
Write	10	4	True	4.759 μs	0.0949 μs	0.2365 μs	4.899 μs	0.1297	-	-	1104 B
Write	10	8	True	4.313 μs	0.0862 μs	0.2031 μs	4.419 μs	0.1297	-	-	1104 B
Write	10	64	True	4.332 μs	0.0861 μs	0.2145 μs	4.478 μs	0.1297	-	-	1104 B
Write	10	256	True	4.422 μs	0.0884 μs	0.2421 μs	4.553 μs	0.1297	-	-	1104 B
Write	100	4	True	30.561 μs	0.6006 μs	1.2928 μs	30.878 μs	0.9766	-	-	8304 B
Write	100	8	True	30.082 μs	0.5921 μs	1.3840 μs	29.725 μs	0.9766	-	-	8304 B
Write	100	64	True	32.152 μs	0.6380 μs	1.5410 μs	31.529 μs	0.9766	-	-	8304 B
Write	100	256	True	36.147 μs	0.7208 μs	1.7269 μs	36.146 μs	0.9766	-	-	8304 B

On the master branch

Method	BatchSize	WriteSize	SequentialKeys	Mean	Error	StdDev	Median	Gen 0	Gen 1	Gen 2	Allocated
Write	1	4	True	1.154 μs	0.0190 μs	0.0233 μs	1.148 μs	0.0362	-	-	304 B
Write	1	8	True	1.146 μs	0.0061 μs	0.0048 μs	1.145 μs	0.0362	-	-	304 B
Write	1	64	True	1.159 μs	0.0095 μs	0.0084 μs	1.158 μs	0.0362	-	-	304 B
Write	1	256	True	1.202 μs	0.0238 μs	0.0377 μs	1.194 μs	0.0362	-	-	304 B
Write	10	4	True	2.295 μs	0.0446 μs	0.0596 μs	2.292 μs	0.0343	-	-	304 B
Write	10	8	True	2.131 μs	0.0426 μs	0.1084 μs	2.111 μs	0.0343	-	-	304 B
Write	10	64	True	2.400 μs	0.0478 μs	0.1217 μs	2.478 μs	0.0343	-	-	304 B
Write	10	256	True	2.078 μs	0.0351 μs	0.0625 μs	2.055 μs	0.0343	-	-	304 B
Write	100	4	True	12.806 μs	0.2391 μs	0.2559 μs	12.724 μs	0.0305	-	-	304 B
Write	100	8	True	12.690 μs	0.2414 μs	0.2371 μs	12.713 μs	0.0305	-	-	304 B
Write	100	64	True	14.839 μs	0.2049 μs	0.1917 μs	14.839 μs	0.0305	-	-	304 B
Write	100	256	True	18.726 μs	0.3654 μs	0.3589 μs	18.770 μs	0.0305	-	-	304 B

[AlgorithmsAreCool]

@CoreyKaylor So i had a bug in the benchmark, I forgot to commit the transaction 😅

But this does show API and P/Invoke overhead nicely