Fix distance threshold bug ScalarInterp

[dragos-ana]

Fixed calculation of normalized weights within the scalar interpolation framework, if the interpolation location is within a distance threshold wrt a reference location.
In that case, the weight of that specific reference point is directly set to 1.0 AND the other weights are reset to 0.0. The previous approach also enabled multiple 1.0 weights, and only the first detected 1.0 was kept.
Unit tests were added to test the bugfix; these tests would have previously failed because the weights would have always been {1.0, 0.0} regardless of the interpolation location 0.0 or 1.0.

[Copilot]

Pull request overview

Fixes scalar interpolation weight normalization when the interpolation location is (nearly) coincident with a reference location by ensuring a single reference gets weight 1.0 and all others are reset to 0.0.

Changes:

• Update calculate_normalized_weights() to reset all weights to 0.0 and stop processing as soon as a reference point is within distance_threshold.
• Remove the post-pass that tried to detect weight == 1.0 after the fact.
• Add unit tests covering the “at reference location” behavior for inverse-distance and exponential weighting.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
src/core/linalg/tests/4C_linalg_utils_scalar_interpolation_test.cpp	Adds new tests that assert exactly one unit weight when interpolating at reference locations.
src/core/linalg/src/dense/4C_linalg_utils_scalar_interpolation.hpp	Clarifies the intended semantics of distance_threshold in the parameter struct.
src/core/linalg/src/dense/4C_linalg_utils_scalar_interpolation.cpp	Implements the corrected “coincident point => one-hot weights” logic inside the weighting loops.