Tests fail on Chrome 150: floating point differences in reprojection results

[offtherailz]

Description

Chrome 150 returns slightly different results (1 ULP) for Math transcendental functions, as allowed by the ECMAScript spec. 4 tests assert exact floating point equality on reprojected coordinates and fail; Chrome 149 passes. Failures look random on CI because GitHub is rolling out runner images with Chrome 150 gradually.

Failing tests:

• test getWMSBoundingBox — web/client/utils/__tests__/CoordinatesUtils-test.js
• should create a request with empty features — web/client/utils/mapinfo/__tests__/arcgis-test.js
• toWMC — web/client/utils/ogc/__tests__/WMC-test.js
• annotationsToGeoJSON with length and area measures — web/client/plugins/Annotations/utils/__tests__/AnnotationsUtils-test.js

Example: miny: 43.70759642778743 (Chrome 150) vs 43.70759642778742 (fixture, Chrome 149).

How to reproduce

• Run npm test on master with Chrome 150

Expected Result

Tests pass on any Chrome version.

Current Result

4 tests fail with last-digit mismatches. Reproduced locally on Chrome 150 with values identical to CI.

• Not browser related

Browser Affected	Version
Chrome	150 (headless)

Other useful information

• Failing job (Chrome 150): https://github.com/geosolutions-it/MapStore2/actions/runs/28945365123/job/85877708818 (PR)
• https://github.com/geosolutions-it/MapStore2/actions/runs/28944532765/job/85874859887 (master)
• Passing job, same run (Chrome 149): https://github.com/geosolutions-it/MapStore2/actions/runs/28945365123/job/85891991691

Fix: compare coordinates with tolerance instead of exact equality. Ready on branch fix_fp_tests_chrome150.