Keep the skill attack rate in dual-wield IAS breakpoints (Double Swing, Dragon Tail with two claws)

[errolgr]

Summary

In the advanced stats panel, the IAS (Frames): line ignores the skill's own attack rate when both hands hold a weapon. That makes Double Swing (always) and Dragon Tail with two claws show wrong breakpoints. This keeps the skill's attack rate in the dual-wield path.

Problem

GetIASBreakpointString adds a skill's attack rate modifier (nAttackRateBonus) to STAT_ATTACKRATE: Double Swing +50 (Skills.txt Param5), Dragon Tail -20 (Param4). When both hands hold weapons, the dual-wield block that applies PD2's faster-weapon rule recomputes the acceleration from the stat alone:

int nBaseIAS = D2COMMON_GetUnitStat(pUnit, STAT_ATTACKRATE, 0);
nAnimAcceleration = nBaseIAS + nMinAccrIAS;
nMinAnimAcceleration = nBaseIAS;

so the modifier is dropped:

• Double Swing needs two weapons, so it always showed Frenzy's breakpoints. Two Hand Axes, 0 IAS: 24 frames shown, 14 actual.
• Dragon Tail with two claws ignored the -20 penalty. Two Katars, 0 IAS: 11 frames shown, 14 actual. With one claw it was already correct.

Changes

File	Change
BH/Drawing/Stats/StatsDisplay.cpp	Dual-wield block: use STAT_ATTACKRATE + nAttackRateBonus, like the single-weapon paths

• One line. Only skills with a non-zero nAttackRateBonus (Double Swing, Dragon Tail) used with two weapons change; every other line is unchanged.

Testing

Against published tables: the expected values come from a model checked against the PD2 wiki Breakpoints page. Dragon Tail row: 14 of 15 cells match exactly. The wiki lists TAS 0 for 16 frames, but the animation reaches 16 frames at -3. Double Swing and Frenzy: the whole sequence equals the wiki's hit 1 + hit 2 - 1 at every TAS from -20 to 200.

In game (PD2 Season 13 client, single player): Barbarian with two Hand Axes, and Assassin with one and with two Katars, at 0 and 30 IAS.

	Result
Before (current BH)	Double Swing 2/2 wrong (Frenzy's numbers), Dragon Tail with two claws 2/2 wrong; controls correct
After (this PR)	all 14 lines correct: 4 fixed + 10 unchanged controls (Dragon Tail with one claw, Frenzy, two-claw attack with and without Burst of Speed, Fanaticism)

Case	Before	After
Double Swing, two Hand Axes, 0 IAS	0 (24) / 3 / 6 / 9 / 14 / 19 / 24 / 32 / 40 / 54 / 70 / 92 / 125 / 180 / 292 / 638	0 (14) / 3 / 13 / 27 / 50 / 89
Dragon Tail, two Katars, 0 IAS	0 (11) / 10 / 26 / 50 / 95	0 (14) / 4 / 11 / 23 / 39 / 63 / 102 / 187
Dragon Tail, one Katar, 0 IAS (unchanged)	0 (14) / 4 / 11 / 23 / 39 / 63 / 102 / 187	same

The clang-cl build and in-game parity gate (10/10 PD2 and 10/10 vanilla launches) pass. The other in-game scenarios are unchanged.

Screenshots

Double Swing, two Hand Axes, 30 IAS

Before	After

Dragon Tail, two Katars, 30 IAS

Before	After

Full before/after table and checked reports that are not bugs

Full evidence (every case, oracle values, provenance, and the reports that turned out not to be BH bugs: single-claw IAS, speed auras/buffs, Cloak of Shadows, the nCurrentFrames TODO, right-click-skill-only): https://github.com/errolgr/BH/blob/evidence/stats-breakpoints/PR-EVIDENCE.md