fix: keep exact snapping, source drawing data and adaptive options layout

Resolve curve intersections through cadkernel, retain empty exact results, gather analytic candidates independently of tessellation, and bound material uploads. Preserve large MINSERT data while limiting rendering work; remove obsolete review narratives.

src/scene/selection.rs

@@ -37,12 +37,6 @@ impl Scene {
         })
     }
 
-    /// The handles plus their leader/annotation partners, in the order given.
-    ///
-    /// Kept separate from the sorted, deduplicated form because a bulk
-    /// selection has to preserve pick order — `selected_handles_in_order`
-    /// reports it — and because sorting per handle is what made selecting
-    /// 186 000 entities one call at a time expensive.
     fn expanded_with_leaders(&self, handles: &[Handle]) -> Vec<Handle> {
         let leaders = self.leaders_by_annotation();
         let mut expanded = Vec::with_capacity(handles.len());
@@ -62,11 +56,6 @@ impl Scene {
         expanded
     }
 
-    /// Select many handles in one pass.
-    ///
-    /// The single-handle path allocates, sorts and deduplicates for every
-    /// entity, and a box selection called it once per entity — most of the
-    /// remaining 786 ms of a 186 460-entity selection.
     pub fn select_entities(&mut self, handles: &[Handle]) {
         if handles.is_empty() {
             return;
@@ -84,10 +73,6 @@ impl Scene {
         }
     }
 
-    /// Deselect many handles in one pass.
-    ///
-    /// `deselect_entity` rebuilds `selected_order` with a `retain` per handle,
-    /// so removing N of M selected entities walked the order list N times.
     pub fn deselect_entities(&mut self, handles: &[Handle]) {
         if handles.is_empty() {
             return;
@@ -1082,10 +1067,6 @@ impl Scene {
 mod tests {
     use super::*;
 
-    /// A box selection used to call `select_entity` once per entity. The bulk
-    /// path has to land on the same selection, in the same pick order —
-    /// `selected_handles_in_order` is read by commands such as DIM, where the
-    /// order the user picked things in decides what they mean.
     #[test]
     fn bulk_selection_matches_selecting_one_at_a_time() {
         use acadrust::entities::Line;
@@ -1141,11 +1122,6 @@ mod tests {
         );
     }
 
-    /// Selecting an entity pulls in any LEADER pointing at it, and that was
-    /// answered by walking the whole document once per selected handle —
-    /// selecting 471 583 entities in a 585 786-entity drawing did not finish
-    /// inside eight minutes. The reverse index has to give exactly what the
-    /// walk gave.
     #[test]
     fn the_leader_index_matches_a_document_walk() {
         use acadrust::entities::Leader;
@@ -1207,10 +1183,6 @@ mod tests {
         );
     }
 
-    /// The reason the cache carries a set as well as a list: drawing into a
-    /// drawing that already has that type must not rebuild either, because the
-    /// rebuild walks every entity the layout owns — 444 937 of them on the
-    /// reproducer, ~100 ms, on every single drawn point.
     #[test]
     fn adding_a_type_already_present_reuses_the_list() {
         use acadrust::entities::{Circle, EntityType, Line};