Update on "[Executorch] Add quantized kv cache to oss ci"

Fixes to make sure quantized kv cache works in oss

Differential Revision: [D66269487](https://our.internmc.facebook.com/intern/diff/D66269487/)

[ghstack-poisoned]

Author: kimishpatel

kernels/quantized/__init__.py

@@ -9,7 +9,7 @@
 try:
     import glob
 
-    import torch
+    import torch as _torch
     import executorch
 
     # Ideally package is installed in only one location but usage of
@@ -23,7 +23,8 @@
     )
     assert len(libs) == 1, f"Expected 1 library but got {len(libs)}"
     logging.info(f"Loading custom ops library: {libs[0]}")
-    torch.ops.load_library(libs[0])
+    _torch.ops.load_library(libs[0])
+    del _torch
 except:
     import logging