#1306 The dms per-element operators split in parallel within a tile: new times dms-polygon-operators: since 20.24.0 dms_polygon, dms_split_polygon and the two convex splits compute the polygons of a tile in parallel; a domain of one tile was cleaned or split on one thread before (the 14,729 CBS buurten: 3.2 s, now 0.63 s on 32 logical processors). dms_split_convex_polygon_keilsnoeyink: the timing table measured again (0.05 and 0.24 s on 57,290 building outlines, 0.64 and 4.8 s on the buurten) and the advice with the new difference (0.04 to 0.18 s and 4.2 s longer, still more than the connectivity saves). dms_polygon_connectivity: growing the polygons themselves is 1.2 to 1.8 times faster than going through the convex parts, no longer 2.6 to 3.1, and the split no longer costs more than the whole direct route. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
#1305 dms_split_convex_polygon_keilsnoeyink and dms_convex_polygon_connectivity, and the grouped forms of dms_polygon_connectivity Two new pages, since 20.24.0: - dms_split_convex_polygon_keilsnoeyink: the minimum number of strictly convex parts for a polygon without holes, what falls back to Hertel and Mehlhorn, the measured counts and times, and when it does not pay: in the connectivity of grown parts the split cost more than the fewer parts saved. - dms_convex_polygon_connectivity: the four forms of dms_polygon_connectivity with convex polygons as premise, the exact test per pair, the slivers it finds that the sweep rounds away, and how much it saves (5 to 20%, since finding the candidates takes most of the time). dms_polygon_connectivity: the forms with a group per polygon, no pair within a group and one pair per pair of groups, the direction of a pair of groups with one and with two sets, and the measured advice to grow the polygons themselves when only their pairs are needed. dms_split_convex_polygon, dms polygon operators and polygon operators: the new operators listed and linked; the testcase description covers both split operators. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>