Pinning the bytes around the invalid one to 0x80 lost coverage: with
error_handler_t::ignore/replace, the serializer re-reads the invalid
byte and decodes the following bytes, and its decoder distinguishes
the continuation classes 0x80-0x8F, 0x90-0x9F, and 0xA0-0xBF, as do
the lexer's range checks.
Iterate those positions over the first and last byte of each class
within the valid range instead (utils::utf8_continuation_bytes). The
invalid byte still takes all 256 values. Defining
JSON_TEST_UTF8_EXHAUSTIVE restores the full Cartesian product, with
the same assertion counts as before #5418.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Niels Lohmann
d304501fa7