Data is lost when parsing large files.
I have prepared 5 test files for different cases.
With their help, I learned that losses are not accidental.
In example.xml, when going to iteration 717 (i = 717), the data is lost.

In the rest of the files, I learned that data loss occurs when the number of characters changes. It looks like some kind of buffer overflow.
In example5.xml I am using randomly generated data using a generator.py.

Several xml files have been prepared to show that this is not an error in the input data, but a problem in the library itself.

I tried to trace the cause of the occurrence, and came to the conclusion that the bug lies in the compiled file.

In the ElementTree.py library file, the line
"events = self._events_queue" 
returns an empty list. This can be seen at iteration 717 in example.xml.

The file example5.xml could not be loaded, so to generate it, run the generator.py.