feat: add support for ARM and Intel architectures in builds and workflows

[DecSmith42]

• Enhanced platform matrix to include ARM architectures (windows-11-arm, ubuntu-24.04-arm, macos-15-intel) across validation and build workflows.
• Enabled runtime identifiers for ARM64 in csproj configuration.
• Improved cleaning process with verbose logging, parallel deletion, and detailed error handling.
• Updated README and renamed tool package for clarity in usage and installation instructions.

[Copilot]

Pull request overview

Adds ARM/Intel architecture support across CI builds and the ArtifactClean tool packaging, while enhancing ArtifactClean’s cleaning implementation and user-facing docs.

Changes:

• Expanded GitHub Actions runner matrices to include Windows/Ubuntu ARM and macOS Intel variants.
• Updated ArtifactClean packaging to include ARM64 runtime identifiers and adjusted AOT-related ILCompiler settings.
• Enhanced ArtifactClean cleaning logic with verbose mode, parallel deletions, and improved traversal behavior; updated README installation instructions.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
README.md	Updates installation package ID and improves usage formatting for artclean.
Invex.Tools.ArtifactClean/Invex.Tools.ArtifactClean.csproj	Adds ARM64 runtime identifiers and tweaks publish/AOT compiler settings.
Invex.Tools.ArtifactClean/Commands.cs	Adds --verbose, refactors dotnet command execution, and parallelizes bin/obj deletion.
.github/workflows/Validate.yml	Expands validation matrix to include additional ARM/Intel runners.
.github/workflows/Build.yml	Expands build matrix and downloads additional platform artifacts in release/publish jobs.
_atom/Build.cs	Updates Atom platform matrix to include new runner labels.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 4 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

process.Start() can throw (e.g., missing dotnet on PATH, invalid working directory, or start failures). Previously the code handled a null process; now this will crash the tool. Wrap the start + read initialization in a try/catch (e.g., for Win32Exception/InvalidOperationException) and emit a clear error to stderr before returning a non-zero outcome (or equivalent).

Suggested change

	process.Start();
	// Should always drain output streams to prevent buffer deadlock
	// When RedirectStandardOutput/Error = true but streams aren't consumed,
	// the process buffer can fill and hang. BeginOutputReadLine() drains asynchronously.
	process.BeginOutputReadLine();
	process.BeginErrorReadLine();
	try
	{
	process.Start();
	// Should always drain output streams to prevent buffer deadlock
	// When RedirectStandardOutput/Error = true but streams aren't consumed,
	// the process buffer can fill and hang. BeginOutputReadLine() drains asynchronously.
	process.BeginOutputReadLine();
	process.BeginErrorReadLine();
	}
	catch (System.ComponentModel.Win32Exception ex)
	{
	Console.Error.WriteLine($"Failed to start 'dotnet {command}' in '{path}': {ex.Message}");
	return;
	}
	catch (InvalidOperationException ex)
	{
	Console.Error.WriteLine($"Failed to initialize output capture for 'dotnet {command}' in '{path}': {ex.Message}");
	return;
	}

[Copilot]

With asynchronous BeginOutputReadLine/BeginErrorReadLine, a single WaitForExit() can return before all output events are fully drained/processed, potentially truncating logs (especially in verbose mode). Consider ensuring async reads complete (e.g., a second WaitForExit() or equivalent pattern) before returning.

Suggested change

	// Ensure asynchronous output/error events are fully drained before returning.
	process.WaitForExit();

[Copilot]

Materializing all subdirectories into an array at every recursion level can increase memory pressure and latency for large repos. You can avoid ToArray() by building two lists during a single enumeration pass (e.g., one for bin/obj deletion and one for recursion), which keeps streaming behavior and still supports the two-pass logic.

[Copilot]

These download steps are highly repetitive and will keep growing as the matrix expands. With actions/download-artifact@v4, consider downloading via a single step using a pattern (e.g., Invex.Tools.ArtifactClean-*) and merge-multiple: true, or drive downloads via a small matrix/loop, to reduce duplication and the chance of missing future runner additions.