HighRiseTPU: split each segment's load with its height within the story

[concrete-sangminlee]

Fixes #435

addForcesFace() in modules/createEVENT/HighRiseTPU/HighRiseTPU.cpp split each mesh segment's force between the floor above and the floor below with locY, the segment's absolute height, instead of its height within the current story. For every story above the first this gives Rabove > A and Rbelow < 0; the total is conserved, but most of the mean load ends up on the roof (about 99 % for TPU model T111_4 at 0°, see #435).

This change uses the height within story i for the split and keeps locY (absolute) for findNearestTAP(). It is algebraically the same correction that LowRiseTPU.cpp received in 5db635e (lines 685–688 there).

Checked on a build of 12b6f99 with this hunk (the tested build differs from this branch only in the wording of the comment):

• only the HighRiseTPU binary differs between the two installed trees, and weuq-0001 gives a byte-identical dakotaTab.out;
• T111_4 at 0° (public record T111_4_000_1.mat), 26 stories: every floor is loaded, the roof gets about half a typical floor, and the total mean load is unchanged (11,439 kN);
• a synthetic tap file with one tap per face and uniform pressure gives h·L·q on each floor and h·L·q/2 on the roof, as expected;
• 38 TPU high-rise records: the patched build agrees with an independent implementation within ±10 % on 36 (details in HighRiseTPU: addForcesFace() splits each story's load with the absolute height locY instead of the height within the story, so most of the mean wind load lands on the roof #435).

🤖 Generated with Claude Code