perf(Charts): Reduce allocations and optimize hot paths in Chart control

[PaulAndersonS]

Description of Change

This PR implements 10 targeted performance optimizations in the Chart control to reduce allocations and improve rendering performance in hot paths:

1. Replace O(n²) IndexOf with Dictionary lookup (CategoryAxis.cs) — distinctXValues.IndexOf(val) inside loops was O(n) per call. A pre-built Dictionary<string, int> gives O(1) lookups.

2. Cache Dictionary.Values.ToList() outside loops (CartesianChartArea.cs) — SideBySideSeriesPosition.Values.ToList()[i] was allocating a new list every iteration. Now cached once before the loop.

3. Cache GetType() results (CartesianChartArea.cs) — Repeated stackingSeries.GetType() and .Name calls replaced with local variables to avoid repeated reflection.

4. Replace List.Contains() with HashSet (CategoryAxis.cs) — groupingValues.Contains(xValues[j]) was O(n). A parallel HashSet<string> provides O(1) membership tests.

5. Replace string concatenation with StringBuilder (CategoryAxis.cs) — String += in a loop in GetLabelContent replaced with StringBuilder to avoid repeated allocations.

6. Manual loop instead of LINQ in stacking calculation (CartesianChartArea.cs) — The GetYValue method already uses an optimized manual loop pattern (verified existing implementation).

7. Single-pass min/max calculation (ErrorBarSegment.cs) — Four separate LINQ iterations (Where().Min(), Where().Max()) replaced with a single loop computing all four values.

8. Reuse list allocations in TrackballBehavior (ChartTrackballBehavior.cs) — new List<T>(...) on every mouse move replaced with clear-and-reuse pattern to reduce GC pressure.

9. Use string.Concat instead of += (RangeColumnSeries.cs) — Avoids intermediate string allocation in tooltip text construction.

10. Replace Cast<T>() with pattern matching (CategoryAxis.cs, CartesianChartArea.cs, ChartTrackballBehavior.cs) — Avoids allocating a LINQ enumerator wrapper; direct iteration with is not T pattern is allocation-free.

Performance Impact

These changes target allocation-heavy hot paths that execute during chart rendering, data grouping, trackball interaction, and tooltip display. The optimizations reduce:

• Unnecessary list/string allocations per frame
• O(n²) algorithms to O(n) with dictionary/hashset lookups
• Repeated reflection calls via caching
• LINQ enumerator allocations via direct iteration

Issues Fixed

N/A — Performance improvement, no functional changes.