PartialFillExchange: remove a resting order that is filled for exactly its leaves_qty

[rux-eth]

What happens

In PartialFillExchange, when a trade fills a resting order for exactly its remaining
quantity, check_if_buy_filled / check_if_sell_filled take the else branch of

let exec_qty = if filled_qty > order.leaves_qty {
    self.filled_orders.push(order.order_id);
    order.leaves_qty
} else {
    filled_qty
};

(hftbacktest/src/backtest/proc/partialfillexchange.rs, two sites). fill() then sets
leaves_qty to 0 and the status to Filled, but the order is never pushed to filled_orders,
so remove_filled_orders() leaves it in self.orders and in its price-level set. The next event
that matches the level — another trade at the price, or a best-price sweep in
on_best_bid_update / on_best_ask_update — calls fill() on a Filled order, which returns
BacktestError::InvalidOrderStatus, and the elapse aborts (return code 14 through the Python
binding). The backtest cannot be resumed past that event.

Fix

Use >= at both sites so an exact fill is removed like an over-fill. Two characters; no other
behaviour changes (an exact fill already produced the Filled status and the response).

Reproduction

Real Hyperliquid L2 + tape data (2025-10-10), HashMapMarketDepthBacktest, RiskAdverseQueueModel,
PartialFillExchange, constant order latency: with one-lot post-only orders across 182 assets the
abort fires about once per 20,000 orders. A minimal synthetic case: a resting bid of 1.0 at a price
with 2.0 displayed ahead of it, then a sell trade of 3.0 at that price (fills the 2.0 ahead and
exactly the 1.0), then any later trade at the same price → InvalidOrderStatus.

Notes

• Found while building a research harness on 2.4.4 (py-v2.4.4); the fix branch is one commit on
  top of that tag. Happy to rebase onto master.
• The Filled-status guard in fill() is what surfaces the stale order; the guard is right, the
  bookkeeping before it is what this PR corrects.

[codacy-production]

Up to standards ✅

🟢 Issues 0 issues

Results:
0 new issues

View in Codacy

🟢 Metrics 0 complexity · -17 duplication

Metric	Results
Complexity	0
Duplication	-17

View in Codacy

_{NEW Get contextual insights on your PRs based on Codacy's metrics, along with PR and Jira context, without leaving GitHub. Enable AI reviewer
TIP This summary will be updated as you push new changes.}