[bug] Add unit test to detect memory leak from data_oriented classes (#…

…6278)

Issue: #6133

### Brief Summary

Co-authored-by: Yi Xu <xy_xuyi@foxmail.com>

requirements_test.txt

@@ -4,6 +4,7 @@ pytest-xdist
 pytest-rerunfailures
 pytest-cov
 numpy
+psutil
 autograd
 requests==2.26
 matplotlib

tests/python/test_memory.py

@@ -1,3 +1,9 @@
+import gc
+import os
+
+import psutil
+from taichi.lang.misc import get_host_arch_list
+
 import taichi as ti
 from tests import test_utils
 
@@ -8,3 +14,36 @@ def test_memory_allocate():
     x = ti.field(ti.i32, shape=(HUGE_SIZE, ))
     for i in range(10):
         x[i] = i
+
+
+@test_utils.test(arch=get_host_arch_list())
+def test_oop_memory_leak():
+    @ti.data_oriented
+    class X:
+        def __init__(self):
+            self.py_l = [
+                0
+            ] * 5242880  # a list containing 5M integers (5 * 2^20)
+
+        @ti.kernel
+        def run(self):
+            for i in range(1):
+                pass
+
+    def get_process_memory():
+        process = psutil.Process(os.getpid())
+        mem_info = process.memory_info()
+        return mem_info.rss / 1e6  # in MB
+
+    # Init & Warm up
+    for i in range(2):
+        X().run()
+        gc.collect()
+
+    ref_mem = get_process_memory()
+    for i in range(5):
+        X().run()
+        gc.collect()
+        curr_mem = get_process_memory()
+        assert (curr_mem - ref_mem < 1E-1
+                )  # shouldn't increase more than 0.1 MB each loop