Hi, I've noticed that after the recent HTML spec changes regarding ProcessingInstructions there are some inconsistencies regarding the tree construction tests (previously at html5lib-tests/tree_construction, now at wpt/html/syntax/parsing/resources) and html5lib's tokenizer tests.

• The tokenizer-subtests "Simili processing instruction" and "A bogus comment stops at >, even if preceded by two dashes" in test2.test, as well as all the tests for the format "<?" followed by a single alpha-char in test3.test no longer have the correct expectation (they expect a (bogus) comment but it should be a processing instruction now)
• All tree construction tests in processing_instructions.dat are missing an "#errors"-section, the explicitly README requires that section ("Then there must be a line that says "#errors"."). It might be a better solution to update the README instead of the tests but one or the other should be updated.
• Some tests in processing_instructions.dat escape special characters in #data (e.g. "\t" instead of the tab character). All other tree construction tests use the raw characters and again according to the README there should be no (un-)escaping happening ("All subsequent lines until a line that says "#errors" are the test data and must be passed to the system being tested unchanged, except with the final newline (on the last line) removed."). (Un-)escaping the input might be an improvement over the current behaviour (e.g. the input becomes more readable) but the current state is inconsistent.