[najlkin]

This PR fixes passing of settings from the old data state to a new one. Among other things, it fixes the point line in a more systematic way than #378 . It also fixes passing of mesh attributes as mentioned in #383 .

[tzanio]

🤖 Suggestions from an AI review. Feel free to ignore if not relevant.

• A pointline command followed immediately by new solution data can lose the line: CloneEmpty() can copy the old coordinates before the queued POINT_LINE command executes. The wait in NewMeshAndSolution() happens after that copy. Could we preserve point coordinates on the main thread when applying the update, as in Retain point line when new data is sent #378, and avoid copying the live point-coordinate vector from the communication thread?

• There may also be a data race: the worker’s CloneEmpty() copies point_coords while the main thread can replace that vector when executing POINT_LINE or installing new data. Could we avoid reading the live vector from the worker, or synchronize those accesses? Copying the coordinates again on the main thread would fix the stale-value issue but leave this concurrent access.