remove questionable vector search optimizations

new round of benchmarks didn't reveal any benefits from them,
quite the opposite.

these were old optimizations added in the early phase of vector
search development. apparently later changes made them obsolete.

Author: vuvova

sql/vector_mhnsw.cc

@@ -28,7 +28,7 @@
 static constexpr float NEAREST = -1.0f;
 
 // Algorithm parameters
-static constexpr float alpha = 1.1f;
+static constexpr float leniency= 1.1f;
 static constexpr uint ef_construction= 10;
 static constexpr uint max_ef= 10000;
 static constexpr size_t subdist_part= 192;
@@ -1179,11 +1179,11 @@ static int select_neighbors(MHNSW_param *p, FVectorNode *target,
   {
     Visited *vec= pq.pop();
     FVectorNode * const node= vec->node;
-    const float target_dista= std::max(32*FLT_EPSILON, vec->distance_to_target / alpha);
+    const float target_dista= std::max(32*FLT_EPSILON, vec->distance_to_target);
     bool discard= false;
     for (size_t i=0; i < neighbors.num; i++)
       if ((discard= node->distance_greater_than(neighbors.links[i]->vec,
-                            target_dista, p->mode, &p->acc) <= target_dista))
+                            target_dista, p->mode, &p->acc) < target_dista))
         break;
     if (!discard)
       target->push_neighbor(p->layer, node);
@@ -1294,7 +1294,6 @@ static int search_layer(MHNSW_param *p, const FVector *target, float threshold,
   Queue<Visited> candidates, best;
   bool skip_deleted;
   uint ef= result_size;
-  const float leniency= 1.1f + p->ctx->M/500.0f;
 
   if (construction)
   {