Urls could be shortened by sending filter data in req.body

[iritmaximus]

http://localhost:5173/crosssearch?&columnfilters=[{%22id%22:%22max_age%22,%22value%22:[%221%22,%2220%22]},{%22id%22:%22min_age%22,%22value%22:[%221%22,%2220%22]},{%22id%22:%22dec_lat%22,%22value%22:[%221%22,%22100%22]},{%22id%22:%22dec_long%22,%22value%22:[%221%22,%22100%22]},{%22id%22:%22lid%22,%22value%22:%221%22},{%22id%22:%22loc_name%22,%22value%22:%22C%22},{%22id%22:%22country%22,%22value%22:%22S%22},{%22id%22:%22state%22,%22value%22:%22C%22},{%22id%22:%22county%22,%22value%22:%22B%22},{%22id%22:%22bfa_max%22,%22value%22:%22m%22},{%22id%22:%22bfa_min%22,%22value%22:%22m%22},{%22id%22:%22frac_max%22,%22value%22:%222%22},{%22id%22:%22frac_min%22,%22value%22:%222%22},{%22id%22:%22basin%22,%22value%22:%22v%22},{%22id%22:%22dms_lat%22,%22value%22:%224%22},{%22id%22:%22dms_long%22,%22value%22:%221%22},{%22id%22:%22site_area%22,%22value%22:%225%22},{%22id%22:%22gen_loc%22,%22value%22:%22n%22},{%22id%22:%22formation%22,%22value%22:%22L%22},{%22id%22:%22member%22,%22value%22:%22C%22},{%22id%22:%22species_id%22,%22value%22:%222%22},{%22id%22:%22subclass_or_superorder_name%22,%22value%22:%22E%22},{%22id%22:%22order_name%22,%22value%22:%22C%22},{%22id%22:%22family_name%22,%22value%22:%22E%22},{%22id%22:%22genus_name%22,%22value%22:%22E%22},{%22id%22:%22species_name%22,%22value%22:%22E%22},{%22id%22:%22body_mass%22,%22value%22:%220%22}]&sorting=[{%22id%22:%22bfa_min%22,%22desc%22:false}]&pagination={%22pageIndex%22:0,%22pageSize%22:15}

[karilint]

Starting implementation. I will inspect the current cross-search table/API request path, add POST body support for large filter/sorting/pagination payloads while preserving existing GET behavior where practical, and validate with focused lint/type checks plus any relevant tests I can run safely.

[karilint]

Implemented in draft PR #1160. Cross-search now has POST body endpoints for table rows, map localities, and CSV export, and the frontend uses those JSON body requests so large filters/sorting/pagination payloads no longer need to be packed into the request URL. The old GET endpoints remain in place for compatibility.

[karilint]

Manual test checklist:\n\n- [ ] Open /crosssearch and apply several column filters, including enough filters to previously make the URL very long.\n- [ ] Confirm the browser URL no longer grows with the cross-search filter payload while the table still updates.\n- [ ] Confirm sorting and pagination still update the table results.\n- [ ] Confirm the locality map data still follows the active cross-search filters.\n- [ ] Use Export table from /crosssearch and confirm the CSV download still uses the active filters and sorting.

[karilint]

Updated manual test checklist:

• Open /crosssearch and apply several column filters, sorting, and pagination; confirm the browser URL contains only tableState and does not grow with columnfilters/sorting/pagination JSON.
• Repeat the same check on another list view such as /locality, /species, /reference, or /museum.
• Refresh the page while the short tableState URL is present and confirm the table state is restored in the same browser tab/session.
• Open an old-style URL containing columnfilters/sorting/pagination and confirm it still loads the table state.
• From a filtered list, open a row detail and use the return-to-table flow; confirm it returns to the short tableState URL and restores the list state.
• Use Export table from /crosssearch and confirm the CSV download still uses the active filters and sorting.

[karilint]

Follow-up after review: the first pass only moved the cross-search data/export API requests to POST bodies. I have now added the shared list-view part too: TableView stores filter/sorting/pagination state in sessionStorage and keeps the browser URL short with a tableState id for all normal table/list views. Legacy long URLs still load as a fallback.

[karilint]

Second follow-up after manual testing: refresh and return-to-table were restoring the short tableState URL but then TableView immediately saved default state over the stored state before restoration finished. I added a load-complete guard so saving is skipped until stored/legacy URL state has been applied, and added a regression test for restoring state from a short tableState URL without overwriting it.