Pivot revenue into a column per category

The monthly trading report wants one row per month and one column per product category, whether or not that category sold anything.

Input

sales sale_id bigint category string -- 'BOOKS', 'GAMES', 'TOYS', and occasionally others sold_at string -- ISO-8601 UTC, 'YYYY-MM-DDTHH:MM:SSZ' amount double -- null until the payment provider confirms

Output

One row per month that has at least one sale, with the columns:

month — 'YYYY-MM'
BOOKS, GAMES, TOYS — that month's revenue in each, rounded to 2 decimal places, and 0.0 where there was none

Categories outside those three are not reported. A month that contains only such sales still gets a row, with three zeros.

Example

month	BOOKS	GAMES	TOYS
2024-01	100.0	20.0	50.5
2024-02	15.25	0.0	0.0

Constraints

The three category columns are fixed. Do not let the data decide the schema — a report whose columns change when a category has a quiet month is not a report.
A null amount contributes nothing. It does not make the month null.
Row order is not graded. Column names and their order are.

Community-reported interview topic. Not an official company question and no affiliation is implied.