Solve this with a window function

You're about to write a self-join and suspect there's a cleaner way.

I want to do the following with a window function rather than a self-join or correlated subquery.

1. Say which window function fits: ROW_NUMBER, RANK, DENSE_RANK, LAG, LEAD, SUM or COUNT over a frame, FIRST_VALUE, NTH_VALUE. Explain why that one and not the neighbouring one, especially ROW_NUMBER against RANK where ties exist.
2. Write the query with an explicit PARTITION BY and ORDER BY, and an explicit frame clause. Do not rely on the default frame, state it and explain what it means.
3. Say what happens at partition boundaries, and what LAG or LEAD returns on the first or last row, plus the default I should supply.
4. Note whether the result changes if the ORDER BY column has ties or NULLs, and how NULLs sort in my dialect.

If a window function is the wrong tool here, say so and show the better approach.

What I want:
Schema:
Dialect:

Writing and reading queries. Drafting SQL from a plain question, making an inherited query legible, and moving it between dialects.