Fix scattermap selection across antimeridian

[coyaSONG]

Overview

Fixes #7904.

Scattermap points are now projected into the world copy nearest the current map center before selection hit-testing. This keeps points on both sides of the antimeridian in the same visible coordinate frame while preserving the existing behavior when world copies are disabled.

The regression test reproduces the reported six-point example and verifies that both box and lasso selection:

• select all six visible points;
• report a continuous box longitude range; and
• report a continuous lasso path across 180 degrees.

Root cause and user impact

MapLibre renders repeated world copies, but scattermap selection normalized every point to [-180, 180] before projecting it. Around the antimeridian, points from one side were therefore projected one full world away from the visible copy and failed the pixel-space selection test. Choosing the longitude copy nearest the map center aligns hit-testing with what the user sees.

Validation

• npm run pretest
• npm run test-jasmine -- select --nowatch (117/117 passing)
• npm run test-syntax
• npm run lint
• git diff --check

The regression test was also run before the production change and failed with only 3 of 6 points selected for both box and lasso selection.

AI assistance

This contribution was prepared with OpenAI Codex assistance. Codex reviewed the repository guidelines, implementation, test coverage, and final diff; no human review is claimed.

[camdecoster]

Thanks for the PR! I'll take a look and follow up with you.