Optimize weighted_choice

This should have no noticeable impact right now because the weight values are small anyway, but for huge weight values this should be a significant optimization as it avoids creating huge lists of duplicates in memory.

randomizer.py

@@ -954,7 +954,7 @@ def get_new_rng(self):
     return rng
 
   def weighted_choice(self, rng: Random, seq: list[T], weight_conditions: list[tuple[int, Callable[[T], bool]]]) -> T:
-    weighted_seq = []
+    element_weights = []
 
     for element in seq:
       weight_for_element = 1
@@ -963,9 +963,9 @@ def weighted_choice(self, rng: Random, seq: list[T], weight_conditions: list[tup
         if condition_callback(element):
           weight_for_element *= weight
 
-      weighted_seq += weight_for_element*[element]
+      element_weights.append(weight_for_element)
 
-    return rng.choice(weighted_seq)
+    return rng.sample(seq, 1, counts=element_weights)[0]
 
   def get_seed_hash(self):
     # Generate some text that will be shown on the name entry screen which has two random character names that vary based on the permalink (so the seed and settings both change it).