test_parse_error for native parser

[lisroach]

I've been taking a look at https://github.com/Instagram/LibCST/issues/1112#issue-2156871432 and struggled a bit getting the tests to reproduce the error, although it was easy to reproduce from a file.

I found this line: https://github.com/Instagram/LibCST/blame/e20e757159f435417fb4d3a0913264c7d252b847/libcst/_parser/tests/test_parse_errors.py#L177 filters out anything other than a check if the error thrown is a SyntaxError when using the native parser, so we lose granularity of testing the error string itself.

Since the errors are different between the native and pure parser, the error messages in this test class will need to be rewritten to match the native output.

A couple of questions:

1. Is my above understanding correct? :) I might have missed if the native parser is tested elsewhere in the codebase.
2. Do we prefer rewriting the file to test just for the output of the native parser, or should we duplicate the tests to test for both native and pure output?