Add Orchard Core command-line evaluation benchmark

[OvesN]

Context

MSBuild has synthetic evaluation benchmarks, but no representative benchmark for targetless -getProperty and -getItem queries against a large SDK-style project. These queries exercise command-line processing, SDK and toolset setup, and project evaluation, and their cost depends on how much of the evaluation pipeline is required to produce the requested result.

Changes Made

Adds a dedicated MSBuild.OrchardCore.Benchmarks executable for benchmarks that depend on an external Orchard Core checkout, keeping those dependencies and configuration separate from the general synthetic benchmark suite.

The OrchardCoreEvaluationBenchmark measures repeated command-line queries against Orchard Core's outer cross-targeting project:

• GetProperty queries TargetFrameworks.
• GetItems queries PackageReference items.
• Each benchmark method executes 100 queries per invocation and reports elapsed time and managed allocations per query.
• Queries use the production MSBuildApp.Execute path, including command-line parsing and evaluation setup, without directly selecting a ProjectEvaluationStage; the code under test determines which evaluation phases run.
• Setup validates that both queries succeed and produce non-empty expected output before measurements begin.

The dedicated runner follows the existing MSBuild.Benchmarks configuration and supports the same ETW, NGEN/ReadyToRun, and JIT-inlining options, while accepting the Orchard Core project path as its workload input.

Testing

• Full repository build
• Dedicated Orchard Core benchmark project build
• Orchard Core benchmark dry run for both query methods

Notes

The benchmark invokes the MSBuild command-line implementation in-process, so it measures repeated query processing and project evaluation but excludes operating-system process startup. It intentionally does not set TargetFramework, matching queries against the outer cross-targeting Orchard Core project. The dedicated project targets the current .NET runtime and requires a restored Orchard Core checkout.

[Copilot]

Pull request overview

Adds an opt-in BenchmarkDotNet benchmark that exercises MSBuild’s -getProperty / -getItem command-line query path against a large, representative SDK-style project (Orchard Core), focusing on evaluation cost and allocations per query.

Changes:

• Added OrchardCoreEvaluationBenchmark to repeatedly execute in-proc MSBuild CLI queries for TargetFrameworks and PackageReference.
• Updated benchmark runner (Program.cs) to include/exclude the Orchard Core benchmark based on a project-path option/env-var.
• Documented how to run the new benchmark and added a new custom CLI option in the benchmarks readme.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
src/MSBuild.Benchmarks/readme.md	Documents how to run the Orchard Core evaluation benchmarks and the new --orchard-core-project option.
src/MSBuild.Benchmarks/Program.cs	Adds argument parsing for --orchard-core-project and conditionally filters benchmark types based on env var presence.
src/MSBuild.Benchmarks/OrchardCoreEvaluationBenchmark.cs	New benchmark measuring repeated -getProperty / -getItem queries via MSBuildApp.Execute, with setup/validation and SDK location logic.
src/MSBuild.Benchmarks/MSBuild.Benchmarks.csproj	Excludes the new benchmark from .NET Framework builds and adds a reference to the MSBuild command-line project.