Perf: Replace O(n²) lookups with Dictionary in CategoryAxis.GroupData

[PaulAndersonS]

Root Cause of the Issue

The CategoryAxis.GroupData() method had two O(n²) performance bottlenecks:

1. List.Contains() in a loop (line 45): For each x-value, groupingValues.Contains() performs a linear scan of the list to check for duplicates — O(n) per call × n calls = O(n²).

2. List.IndexOf() in a LINQ query (lines 73, 77): For each value in a series, distinctXValues.IndexOf(val) performs a linear search — O(n) per call × n calls = O(n²).

For charts with large datasets (thousands of data points), these patterns cause noticeable delays during chart initialization and data grouping.

Description of Change

• Replaced List.Contains() with HashSet.Add() — provides O(1) amortized lookup for deduplication while maintaining insertion order via the parallel list.
• Built a Dictionary<string, int> index lookup — maps each distinct x-value to its index for O(1) lookups instead of O(n) List.IndexOf() calls.
• Used explicit pattern matching for ActualXValues type checks (is List<double> instead of as List<double>).
• Pre-allocated lists with known capacity to reduce reallocations.

Complexity improvement: O(n²) → O(n) for the grouping operation.

Unit Tests Added

5 new unit tests covering:

• Duplicate values produce correct indexes
• Double (numeric) x-values are handled correctly
• Multiple series with overlapping values deduplicate properly
• Single series produces sequential indexes
• Large dataset (1000+ items) completes with correct results

Issues Fixed

N/A — Proactive performance improvement

Screenshots

N/A — No visual changes