FsharpCompilerService benchmark for changes in a cascade of files referencing each other

[T-Gro]

This benchmark suite now proves that F# is not very clever in incremental builds - any change of files invalidates the subsequent files, even if their signatures would not change.

It can be used as a basis for testing improvement via new features/ideas.

The current results are:

BenchmarkDotNet=v0.13.2, OS=Windows 11 (10.0.22000.1098/21H2)
11th Gen Intel Core i9-11950H 2.60GHz, 1 CPU, 16 logical and 8 physical cores
.NET SDK=7.0.100-rc.2.22477.23
  [Host]  : .NET 7.0.0 (7.0.22.47203), X64 RyuJIT AVX2 DEBUG
  LongRun : .NET 7.0.0 (7.0.22.47203), X64 RyuJIT AVX2

Job=LongRun  IterationCount=100  LaunchCount=3  
WarmupCount=15  

Method	Mean	Error	StdDev	Median	Ratio	RatioSD	Rank	Gen0	Gen1	Gen2	Allocated	Alloc Ratio
ParseProjectAsIs	98.19 ns	0.723 ns	3.766 ns	98.70 ns	0.001	0.00	I	0.0223	-	-	280 B	0.17
ParseProjectWithFullCacheClear	89,898.93 ns	190.661 ns	986.896 ns	89,727.05 ns	1.000	0.00	II	1000.0000	1000.0000	1000.0000	1664 B	1.00
ParseProjectWithChangingFirstFile	531,555.27 ns	2,579.920 ns	13,145.366 ns	531,244.04 ns	5.914	0.16	III	-	-	-	8600 B	5.17
ParseProjectWithChangingMiddleFile	544,790.07 ns	7,099.963 ns	36,750.697 ns	553,684.03 ns	6.061	0.41	IV	-	-	-	8896 B	5.35
ParseProjectWithChanging25thPercentileFile	553,487.15 ns	7,892.799 ns	40,713.491 ns	551,352.78 ns	6.158	0.46	IV	-	-	-	8896 B	5.35

[vzarytovskii]

Great! Might be good to eventually include it in dotnet/performance too.

[psfinaki]

This benchmark suite now proves that F# is not very clever in incremental builds