Add delta computed calculation

[jamesmoore]

Attempt to improve calculation performance. No significant gains.

[Copilot]

Pull request overview

Adds a new solver variant (AutogramBytesNoStringsV8) that updates computedCounts incrementally (delta-based) rather than fully recomputing each iteration, and wires it into tests and benchmarks.

Changes:

• Introduce AutogramBytesNoStringsV8 with delta-based computed-count updates and debug verification.
• Add an xUnit test case for V8 to ensure it still finds a valid autogram with the expected iteration count.
• Update benchmarks to include V8 and comment out several older benchmark variants.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 6 comments.

File	Description
AutogramTest/AutogramTest.cs	Adds TestV8() to validate the new solver implementation.
AutogramBenchmark/AutogramBenchmark.cs	Adds V8 benchmark wiring and comments out multiple older benchmark variants.
Autogram/AutogramBytesNoStringsV8.cs	New V8 solver using delta updates for computedCounts.

---

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

[Copilot]

The DEBUG-only verification in DeltaUpdateComputedCounts calls ReCalculateComputedCounts() every iteration, allocating a new array and doing an O(n^2) recompute. With the existing tests running ~1.7M iterations, this can severely slow/debug builds and may cause excessive GC or timeouts. Consider guarding this behind a dedicated compilation symbol, sampling (e.g., every N iterations), or reusing a buffer to avoid per-iteration allocations.

[Copilot]

This benchmark class now has several solver list fields (e.g., V5a/V5b/V6/V7) that are no longer referenced because the corresponding setup/cleanup/benchmark methods are commented out. These become dead code and can introduce warnings/confusion; consider removing the unused fields or re-enabling the benchmarks via BenchmarkDotNet categories/filters instead of commenting them out.

Suggested change

private List<AutogramBytesNoStringsV8> solver8List = null!;

[Copilot]

Large commented-out IterationSetup blocks make it harder to tell which benchmarks are intended to run and are easy to forget to re-enable. Prefer deleting unused benchmark variants, or using BenchmarkDotNet mechanisms (categories, filters, or config) / conditional compilation to control which benchmarks execute instead of leaving commented code in the source.

Suggested change

	//[IterationSetup(Targets = new[] { nameof(AutogramBytesNoStringsV4_Solve_Average_Batched_Seeds) })]
	//public void IterationSetupV4()
	//{
	// solver4List = CreateSolvers(SeedCount, p => new AutogramBytesNoStringsV4(autogramConfig, p));
	//}

[Copilot]

The expression 'A == false' can be simplified to '!A'.

Suggested change

	var wasReordered = computedCounts.AsSpan().SequenceEqual(proposedCounts) == false;
	var wasReordered = !computedCounts.AsSpan().SequenceEqual(proposedCounts);

[Copilot]

The expression 'A == false' can be simplified to '!A'.

Suggested change

	if (history.Contains(result) == false)
	if (!history.Contains(result))