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/io/mod.rs

@@ -744,18 +744,6 @@ pub fn load_bytes(name: &str, bytes: Vec<u8>) -> Result<CadDocument, String> {
     }
 }
 
-/// [`load_bytes`] plus the same finalization and corruption handling the
-/// canonical path-based open (`load_file_for_open`/`finalize_loaded_outcome`)
-/// already runs — for a byte-based open that nonetheless has a real
-/// filesystem path behind it (automation's `"open"` op reads the file into
-/// memory itself, e.g. to go through an edit lease, but the drawing still
-/// lives at `path`). Without this, automation-opened documents kept
-/// legacy non-zero block origins un-normalized, left relative raster/material
-/// paths unresolved, carried no `source_path`, and skipped the corrupt-entity
-/// purge the UI open path always runs — silently behaving differently from a
-/// UI open of the exact same file. `load_bytes` itself is unchanged (and
-/// still used directly by the byte round-trip tests, which have no real path
-/// and don't want this finalization).
 #[cfg(not(target_arch = "wasm32"))]
 pub fn load_bytes_finalized(path: &Path, bytes: Vec<u8>) -> Result<(CadDocument, usize), String> {
     let name = path
@@ -2061,17 +2049,7 @@ pub(crate) fn is_entity_corrupt(e: &EntityType) -> bool {
     // rarely use this many — and parser desync produces exactly-100_000-vertex
     // junk records.
     const MAX_VERTS: usize = 100_000;
-    // A MINSERT's row/column counts are parsed `u16`s whose unchecked product
-    // (`Insert::instance_count`) drives a per-instance allocation in
-    // `render_graph::array_offsets` and a full per-instance render-graph walk
-    // for every one — a corrupt or adversarial 65535x65535 pair requests over
-    // four billion instances (tens of GiB) long before any GPU-side chunking
-    // limit could help. Kept in sync with the identical failsafe clamp in
-    // `render_graph::array_offsets`, which guards inserts that reach the
-    // render graph some other way (e.g. built at runtime, not loaded).
-    const MAX_MINSERT_INSTANCES: usize = 20_000;
     match e {
-        E::Insert(i) => i.instance_count() > MAX_MINSERT_INSTANCES,
         E::LwPolyline(p) => {
             !finite_unit_normal(&p.normal)
                 || p.vertices.len() >= MAX_VERTS
@@ -2451,11 +2429,11 @@ mod corrupt_guard_tests {
     // product can reach into the billions) must be rejected before it can
     // drive the render graph's per-instance allocation and expansion.
     #[test]
-    fn rejects_pathological_minsert_counts() {
+    fn preserves_large_minsert_data_while_rendering_is_bounded() {
         let mut i = acadrust::entities::Insert::new("BLOCK", Vector3::ZERO);
         i.row_count = u16::MAX;
         i.column_count = u16::MAX;
-        assert!(is_entity_corrupt(&EntityType::Insert(i)));
+        assert!(!is_entity_corrupt(&EntityType::Insert(i)));
     }
 
     // An ordinary array insert, well under the budget, is valid source data.