Flaky sdk test on pypy

=================================== FAILURES ===================================
_ TestBatchProcessor.test_record_processor_is_garbage_collected[BatchSpanProcessor-telemetry1] _

self = <tests.shared_internal.test_batch_processor.TestBatchProcessor object at 0x000000002da681e0>
batch_processor_class = <class 'opentelemetry.sdk.trace.export.BatchSpanProcessor'>
telemetry = <opentelemetry.sdk.trace.ReadableSpan object at 0x000000002da11600>

    def test_record_processor_is_garbage_collected(self, batch_processor_class, telemetry):
        exporter = Mock()
        processor = batch_processor_class(exporter)
        weak_ref = weakref.ref(processor)
        processor.shutdown()
    
        # When the processor is garbage collected
        del processor
        gc.collect()
    
        # Then the reference to the processor should no longer exist
>       assert weak_ref() is None
E       AssertionError: assert <opentelemetry.sdk.trace.export.BatchSpanProcessor object at 0x0000000031a74368> is None
E        +  where <opentelemetry.sdk.trace.export.BatchSpanProcessor object at 0x0000000031a74368> = <weakref at 0x00007f07dbdf37a0; to 'BatchSpanProcessor'>()

opentelemetry-sdk/tests/shared_internal/test_batch_processor.py:204: AssertionError

Start with opentelemetry-sdk/tests/shared_internal/test_batch_processor.py and test_record_processor_is_garbage_collected; run that test under PyPy and compare the reference behavior with the test's assumptions. Make the test reliably validate processor collection on PyPy, then rerun the focused test and relevant batch-processor tests.