Use both ends of the pipe in parallel

[JesHansen]

By splitting the pipe into the writing end and reading end it seems to be able to slightly boost performance. On my system it benchmarks as roughly ~10% faster to have the file being parsed while it is loading in.

BenchmarkDotNet=v0.13.0, OS=Windows 10.0.19042.985 (20H2/October2020Update)
AMD Ryzen 7 5800X, 1 CPU, 16 logical and 8 physical cores
.NET SDK=5.0.203
  [Host]     : .NET 5.0.6 (5.0.621.22011), X64 RyuJIT
  DefaultJob : .NET 5.0.6 (5.0.621.22011), X64 RyuJIT

Method	Mean	Error	StdDev	Rank	Gen 0	Gen 1	Gen 2	Allocated
FullPipe	51.77 ms	0.363 ms	0.303 ms	1	1000.0000	500.0000	-	17 MB
PipeLines	56.97 ms	0.400 ms	0.355 ms	2	1000.0000	444.4444	-	17 MB
CsvHelper	136.83 ms	1.827 ms	1.709 ms	3	5250.0000	2250.0000	1500.0000	77 MB
AsyncStream	153.04 ms	2.995 ms	4.483 ms	4	4250.0000	2500.0000	750.0000	64 MB