Store each player's leaderboard totals (wins, accumulated points and the
latest display name) on player_rating alongside the Elo rating, updated
in the same transaction as the match, so the board is a single indexed
table query instead of aggregating every participation on each request.
GET /api/leaderboard is now page-number paginated (page/limit) and
returns the page, page count and total, plus the caller's own rank and
page when authenticated; focus=me returns the caller's page directly.
The migration adds the columns, backfills them from the match history
(inserting rows for players from pre-Elo matches at the initial rating)
and replaces the index. backfill_elo rebuilds the aggregates too.
The web page opens on the page containing the signed-in player, marks
their row and adds a numbered footer pager.
Cursors are JSON, so a datetime sort key reached the next request as a
string. Tortoise's DatetimeField.to_db_value does not parse strings, so
asyncpg rejected the string where a timestamptz parameter was expected;
sqlite tolerated it, hiding the bug in tests. Coerce decoded cursor
values through their model field's to_python_value before building the
keyset filter.
woggioni