Optimize Pipe.DefaultPipeWriter GetSpan and Advance

[ahsonkhan]

• Memory.Slice.Span is slightly slower than Memory.Span.Slice (GetSpan helper gets the span first before slicing).
• Removing the second (length) parameter from Slice since it isn't necessary.
• I am assuming that Advance(bytesWritten = 0) is not a common scenario, so we shrink the method size by not creating a separate code path for it and save a branch.

cc @pakrym, @davidfowl, @benaadams

[benaadams]

nit: availableMemory => availableSpan

[pakrym]

Don't use _writingHead outside the lock

[benaadams]

Was used previously outside lock?

[pakrym]

Good time to stop :)

[pakrym]

Or wait, it's not supposed to be syncronised.

[pakrym]

Any perf numbers?

[ahsonkhan]

Any perf numbers?

I don't know of any relevant benchmarks that exist today that will help with this (at this scale). Any ideas which ones to run and compare? Nano benchmarks, measuring each of the methods in isolation show small improvements (~10%).

BenchmarkDotNet=v0.10.14, OS=Windows 10.0.16299.431 (1709/FallCreatorsUpdate/Redstone3)
Intel Xeon CPU E5-1620 v2 3.70GHz, 1 CPU, 8 logical and 4 physical cores
Frequency=3604598 Hz, Resolution=277.4234 ns, Timer=TSC
.NET Core SDK=2.1.300-rtm-008866
  [Host]     : .NET Core 2.1.0-rtm-26516-03 (CoreCLR 4.6.26516.03, CoreFX 4.6.26516.03), 64bit RyuJIT
  DefaultJob : .NET Core 2.1.0-rtm-26516-03 (CoreCLR 4.6.26516.03, CoreFX 4.6.26516.03), 64bit RyuJIT

If marked with AggressiveInlining attribute:

Method	Mean	Error	StdDev	Scaled	ScaledSD
GetMemorySpan	3.728 ns	0.0300 ns	0.0234 ns	1.00	0.00
GetSpanDirectly	3.254 ns	0.0973 ns	0.1121 ns	0.87	0.03

If not inlined:

Method	Mean	Error	StdDev	Scaled
GetMemorySpan	13.275 ns	0.0895 ns	0.0699 ns	1.00
GetSpanDirectly	4.238 ns	0.1177 ns	0.1488 ns	0.32

Method	Mean	Error	StdDev	Scaled	ScaledSD
GetSegmentSizeOriginal	2.926 ns	0.0296 ns	0.0262 ns	1.00	0.00
GetSegmentSizeNew	2.865 ns	0.0571 ns	0.0506 ns	0.98	0.02

Method	Mean	Error	StdDev	Scaled
GetMemoryOriginal	24.42 ns	0.0568 ns	0.0531 ns	1.00
GetMemoryNew	21.67 ns	0.1069 ns	0.0947 ns	0.89