docs(shipwrecks): say the personal-text filter is a keyword filter, not a guarantee that no name is shown

DATA_SOURCES and CURRENT-STATE said text naming people is dropped. The filter
drops text with a title or role word (Mr, captain, owner, crew, reported by);
a bare name with none of them still passes. The docs and the code comment
now say so.

DATA_SOURCES.md

@@ -39,7 +39,7 @@ How to read this:
 | **NASA GIBS** (WMTS tiles + Worldview Snapshots) and **NASA CMR** (granule catalog) | Recent Imagery layer: HLS S30/L30 (Sentinel-2, Landsat 8/9) and VIIRS true-colour tiles draped on the map, per-day thumbnails and PNG exports; the HLS granule catalog (dates, cloud, footprints) for a selected box | U.S. public domain (NASA data); keyless, browser-direct (CORS `*`); no bulk use — one 30-day catalog query per box, tiles bounded to the box, six tile requests in flight | "We acknowledge the use of imagery provided by services from NASA's Global Imagery Browse Services (GIBS), part of NASA's Earth Science Data and Information System (ESDIS)." — [gibs.earthdata.nasa.gov](https://gibs.earthdata.nasa.gov), [HLS](https://lpdaac.usgs.gov/products/hlss30v002/) |
 | **NIFC WFIGS** (Wildland Fire Interagency Geospatial Services, ArcGIS feature service) | Fire Perimeters layer: current interagency wildfire incident perimeters, containment, and incident facts | U.S. public domain (interagency wildland-fire data published through the [NIFC Open Data portal](https://data-nifc.opendata.arcgis.com/)); fetched keyless through `/api/fire-perimeters` (5-minute server cache) with `exceededTransferLimit` paging | "Wildfire perimeters: National Interagency Fire Center (WFIGS)" |
 | **S2 Underground GhostMaps** (`github.com/s2underground/GhostMaps`) | S2 GhostMaps layer: the current edition of S2's KMZ intel products (Common Intelligence Picture, 30-day kinetic pull, Border Crisis Map, Iran and Yemen products, newest incident drop). Keyless. `/api/ghostmaps` reads the branch head from git's ref advertisement every 15 s (`GHOSTMAPS_POLL_SECONDS`, minimum 5 s), lists the tree through the GitHub REST API only when the head changes (optional `GITHUB_TOKEN`), downloads files by commit-pinned raw URLs into `.gev-cache/ghostmaps/` and pushes new catalogs to browsers over server-sent events | [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) (repository README: "S2 Underground © 2025 by S2A1"); non-commercial, attribution required; files are fetched at runtime, never stored in this repository | "S2 GhostMaps intel products: S2 Underground © by S2A1, CC BY-NC-SA 4.0" — shown in-app |
-| **EMODnet Human Activities — Ship wrecks** (`ows.emodnet-humanactivities.eu/wfs`, layer `emodnet:heritageshipwrecks`) | Shipwrecks layer: about 7,000 wrecks in European waters (name, sinking year and text, vessel type, least depth and its quality, position accuracy, period, cargo, protected status). Keyless WFS GeoJSON, fetched by `/api/shipwrecks` once a day; rows sourced from wrecksite.eu are skipped and text naming people is dropped | [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) (EMODnet catalogue record `e965088b-a265-4517-84df-c49b156af8a7`; data originator AND-International, compiled from SHOM, National Monuments Service, Historic England and the Oxford Roman Economy Project); fetched at runtime, not stored in this repository | "Shipwrecks (Europe): EMODnet Human Activities (data originator AND-International, …), CC BY 4.0" — shown in-app |
+| **EMODnet Human Activities — Ship wrecks** (`ows.emodnet-humanactivities.eu/wfs`, layer `emodnet:heritageshipwrecks`) | Shipwrecks layer: about 7,000 wrecks in European waters (name, sinking year and text, vessel type, least depth and its quality, position accuracy, period, cargo, protected status). Keyless WFS GeoJSON, fetched by `/api/shipwrecks` once a day; rows sourced from wrecksite.eu are skipped, and free text that refers to a person by a title or role is dropped by a keyword filter (a bare name can still pass) | [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) (EMODnet catalogue record `e965088b-a265-4517-84df-c49b156af8a7`; data originator AND-International, compiled from SHOM, National Monuments Service, Historic England and the Oxford Roman Economy Project); fetched at runtime, not stored in this repository | "Shipwrecks (Europe): EMODnet Human Activities (data originator AND-International, …), CC BY 4.0" — shown in-app |
 | **Wikidata — shipwrecks** (`query.wikidata.org/sparql`) | Shipwrecks layer: shipwreck items (Q852190) with a coordinate and at least one Wikipedia article, plus sinking events (Q906512, where the death toll lives): name, ship type, sinking date with its precision, deaths and survivors (counts only), description and article link. Keyless, fetched by `/api/shipwrecks` once a day | [CC0](https://creativecommons.org/publicdomain/zero/1.0/); fetched at runtime | "Shipwrecks (worldwide): Wikidata (CC0)" — shown in-app |
 | **InciWeb** (inciweb.wildfire.gov)                                    | Per-incident "InciWeb ↗" card links to official incident pages (narratives, evacuation/closure notices) | U.S. government public incident information. Links are matched best-effort by incident name and state, with incident page origin and update times checked on selection. Same-origin `/api/fire-perimeters/inciweb` routes cache the catalog for 1 hour and publication metadata for 30 minutes; linked pages remain InciWeb content | "Incident information: InciWeb (inciweb.wildfire.gov)" |
 | **FAA Special Use Airspace** (`services6.arcgis.com` FeatureServer `Special_Use_Airspace`) | Prohibited & Restricted Airspace layer: US prohibited (P) and restricted (R) special-use airspace polygons (not the UAS Facility Map or UAS flight restrictions), altitude limits from the `*_VAL`/`*_UOM`/`*_CODE` fields, times of use, and controlling agency | U.S. public domain (FAA aeronautical data published through the [FAA AIS Open Data portal](https://adds-faa.opendata.arcgis.com/)); fetched keyless through `/api/drone-airspace` (1-hour server cache). Informational only — not a substitute for official UAS flight-planning tools | "Prohibited & restricted airspace: FAA Special Use Airspace" |

docs/CURRENT-STATE.md

@@ -166,9 +166,11 @@ its earlier records as stale for up to seven days and is not asked again for
 is not openly licensed). A Wikidata wreck with the same name within 3 km of an
 EMODnet wreck is merged into one record that keeps the article link and takes
 EMODnet's year, sinking text, depth and position accuracy. Only ship columns are
-read: EMODnet's long histories (`ship_char`, `obj_desc`) are not, and any
-sinking or cargo text that names a person (owner, master, a diver who reported
-it) is dropped whole; deaths appear only as a Wikidata count.
+read: EMODnet's long histories (`ship_char`, `obj_desc`) are not. Sinking,
+cargo and description text that refers to a person by a title or role (Mr,
+captain, master, owner, crew, "reported by", …) is dropped whole by a keyword
+filter; a bare name with no such word can still pass. Deaths appear only as a
+Wikidata count.
 
 The globe draws a round marker with a hull and mast per wreck, all in one
 billboard collection, ground-clamped with the depth test off and the horizon

src/layers/shipwrecks/records.js

@@ -45,7 +45,9 @@ const MAX_NOTE = 260;
 const SAME_WRECK_KM = 3;
 
 // Free text sometimes names people (owners, masters, divers who reported a
-// wreck). The layer shows ships, never people, so such text is dropped whole.
+// wreck). The layer is about ships, so text that refers to a person by a title
+// or role is dropped whole. This is a keyword filter: a bare name with none of
+// these words still passes.
 const PERSON_TEXT =
   /\b(mr|mrs|ms|dr|capt|captain|master|skipper|owner|owned|crew|body|bodies|drowned|reported by|according to)\b/i;
 
@@ -60,7 +62,7 @@ function clean(value, max = MAX_TEXT) {
   return text.length > max ? `${text.slice(0, max - 1).trimEnd()}…` : text;
 }
 
-/** Free text with no personal names in it, or null. */
+/** Free text with none of the person keywords in it, or null. */
 function impersonalText(value, max) {
   const text = clean(value, max);
   return text && !PERSON_TEXT.test(text) ? text : null;