Benchmark FlowALP YieldVault rebalancing under simulated market crash

[zhangchiqing]

Problem Definition

The FlowALP and FlowYieldVault rebalancers use scheduled transactions to maintain position health.

During a rapid market movement (e.g., 30%+ price drop), many positions may breach their rebalance thresholds simultaneously. We need to validate:

1. Per-transaction compute: Does a single worst-case rebalance cascade fit within Cadence's per-tx computation limit (9999 gas limit)?
2. Scheduling throughput: When N positions trigger concurrently, can FlowTransactionScheduler queue and execute them without backpressure or dropped schedules?
3. End-to-end latency: From price breach to position rebalanced—what's the p50/p99 latency under load?