Avoid repeated work when displaying long proteins

[jkhales]

Displaying long proteins is slow because each visible row processes the entire protein sequence, including residues it never draws.

This limits non-wrapping translations to the residues in the current row and reuses their click-target IDs when redrawing. A row showing 40 residues from a 600-residue protein now processes 40 instead of 600.

[guzmanvig]

Hi

Thanks for submitting this. Could you share an example video and/or code of the issue you were experiencing and that this PR solves?

Thanks

[jkhales]

Hi @guzmanvig — sure! Recording below. It's from a side-by-side demo of published seqviz@3.10.24 (left) vs a build of this branch (right), both rendering the same sequence.

Load time is marginally faster from the PR, but the difference is most noticeable while scrolling. Same mechanism speeds them both up.
Happy to share the demo page itself (a single HTML file comparing the two builds) if you'd like to run it locally.

Screen.Recording.2026-09-19.at.12.20.59.PM.mov