fix(py-hftbacktest): eliminate per-event Vec allocation in market depth fusion

[caiyi0616]

Summary

Replace per-call Vec<Event> allocation in FusedHashMapMarketDepth update
methods with append-style *_into variants that write into caller-owned storage.

Motivation

Issue #320: The Python market-depth fusion path creates a temporary
Vec<Event> for each accepted depth or snapshot event, causing
approximately one short-lived allocation per accepted input event.

Performance improvement:

Platform	Baseline ns/event	Candidate ns/event	Improvement
Windows 11	58.100	32.124	44.69%
Debian WSL2	41.911	30.116	28.37%

Allocation reduction (1M events):

• Median allocation calls: 980,071.5 → 31
• Cumulative requested bytes: -65.15%

Changes

hftbacktest/src/depth/fuse.rs

Add *_into variants that accept &mut Vec<Event> as output parameter:

• update_bid_depth_into(&mut self, ev: Event, out: &mut Vec<Event>)
• update_ask_depth_into(&mut self, ev: Event, out: &mut Vec<Event>)
• update_best_bid_into(&mut self, ev: Event, out: &mut Vec<Event>)
• update_best_ask_into(&mut self, ev: Event, out: &mut Vec<Event>)

Existing vector-returning methods are retained as thin wrappers that call *_into.

py-hftbacktest/src/fuse.rs

Replace slf.fused.append(&mut evs) with direct *_into calls.
When add=false, pass a ManuallyDrop<Vec> scratch buffer to still
update depth state without recording fused events—preserving the
existing semantics.

Verification

• cargo check for both hftbacktest and py-hftbacktest pass
• All 260 benchmark pairs (from issue Avoid per-event allocation in Python market-depth fusion #320) produce identical output counts and sequence checksums
• Event ordering, timestamps, flags, prices and quantities unchanged

Closes #320

[codacy-production]

Up to standards ✅

🟢 Issues 0 issues

Results:
0 new issues

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.}