perf: remove redundant .lower() calls in Headers

The _list stores (raw_key, lowercase_key, value) tuples, so the
second element is already lowercase. Calling .lower() on it again
is unnecessary and wastes CPU cycles.

Benchmark shows ~1.6x speedup for header lookups.

Author: shindonghwi

httpx/_models.py

@@ -260,7 +260,7 @@ def get_list(self, key: str, split_commas: bool = False) -> list[str]:
         values = [
             item_value.decode(self.encoding)
             for _, item_key, item_value in self._list
-            if item_key.lower() == get_header_key
+            if item_key == get_header_key
         ]
 
         if not split_commas:
@@ -334,7 +334,7 @@ def __delitem__(self, key: str) -> None:
         pop_indexes = [
             idx
             for idx, (_, item_key, _) in enumerate(self._list)
-            if item_key.lower() == del_key
+            if item_key == del_key
         ]
 
         if not pop_indexes: