Build a Report as Text and UTF-8 Bytes

Your shop exports its stock as a small comma-separated report. Write the two functions that render it, one as text and one as UTF-8 bytes. The starter builds both with += in a loop. The bytes version copies the whole report for every line, so it slows down badly on large reports, and its ASCII encoding fails on a name like café. The text version is fast only because of the CPython shortcut that the article describes.

rows = [Row("widget", 3, 1.5), Row("café", 1, 9.999)]
assert render_report(rows) == "name,qty,price\nwidget,3,1.50\ncafé,1,10.00\n"
assert render_bytes(rows) == render_report(rows).encode("utf-8")

What to write

• render_report(rows: Iterable[Row]) -> str appends the lines to a list[str] and joins them once.
• render_bytes(rows: Iterable[Row]) -> bytes adds each UTF-8 encoded line to a bytearray, then converts it to bytes once at the end.

Both use the same format:

• Start with HEADER, which is "name,qty,price\n", even when there are no rows.
• Then write one line per row, in input order, formatted exactly as f"{row.name},{row.qty},{row.price:.2f}\n". Every line ends in a newline.
• Write names as they are, with no quoting or escaping. This is a simple comma-separated format, not a full CSV writer.
• Format prices with .2f, so 9.999 becomes 10.00. Don't round them any other way. Python stores a float such as 12.005 as a value slightly off from what you typed, and .2f rounds the stored value.
• For the same rows, render_bytes returns exactly what render_report(rows).encode("utf-8") returns.

Each function must work on input that can be read only once, such as a generator. Loop over rows once, and give each renderer its own input.

What the driver reports

Keep Row, HEADER, parse and solve as they are. parse(raw, repeat) turns each [name, qty, price] row of strings into a Row, and repeats the whole list repeat times. A repeat of zero or less gives no rows, and the header still appears once. solve renders the same rows both ways and returns:

The largest test renders 200,000 rows. Your submission must pass mypy --strict.