chore: micro & macrobenchmarks

[leonardochaia]

The macrobenchmark does a complete I/O benchmark. It is the same as running dotnet affected command.
It discovers csproj files, creates an MSBuild project graph and then applies the detection algorithms.

The microbenchmark runs the detection algorithm only. The Projects are already discovered and the ProjectGraph is already built.

According to my profiling, the most time consumer method is the building of the ProjectGraph.
Instantiating ProjectGraph for 1k SDK based projects takes ~5s and 3GB ram 😮

BenchmarkDotNet=v0.13.1, OS=manjaro 
11th Gen Intel Core i7-1165G7 2.80GHz, 1 CPU, 8 logical and 4 physical cores
.NET SDK=6.0.200
  [Host]     : .NET 6.0.2 (6.0.222.6406), X64 RyuJIT
  Job-NDREBB : .NET 6.0.2 (6.0.222.6406), X64 RyuJIT

Toolchain=.NET 6.0  WarmupCount=1  

Method	TotalProjects	ChildrenPerProject	Mean	Error	StdDev	Gen 0	Gen 1	Allocated
AffectedAlgorithm	500	20	57.42 μs	0.366 μs	0.324 μs	3.1128	0.1221	19 KB
AffectedAlgorithm	1000	20	925.54 μs	18.312 μs	19.593 μs	-	-	67 KB

Method	TotalProjects	ChildrenPerProject	Mean	Error	StdDev	Gen 0	Gen 1	Gen 2	Allocated
MacroBenchmark	500	20	2.654 s	0.0531 s	0.1261 s	266000.0000	53000.0000	1000.0000	1 GB
MacroBenchmark	1000	20	5.687 s	0.1134 s	0.2185 s	546000.0000	107000.0000	1000.0000	3 GB

Gonna research if there are ways to improve this by configuring the ProjectGraph, but now we have a way to measure it.
We could add a separate benchmark for ProjectGraph instantiation only in order to measure this.

I did notice that if the project has no Sdk defined, it takes significantly less time/resources.
However, we will need to have the Sdk defined in order to use Predictions in PR #46