perf: make priority dequeue 240% faster

[saripovdenis]

Summary

Replaces PriorityQueue#dequeue() Array#shift() with a head cursor and lazy compaction.

This avoids repeatedly moving remaining array entries on every dequeue while preserving queue behavior.

The previous shift() implementation also passes the new behavior tests; the tests verify behavior preservation, while the benchmarks show the performance gain.

Benchmarks

Node.js v20.19.5

Benchmark	Previous implementation	Current implementation	Change
large-fifo-backlog	48.32 ops/sec ±6.96%	58.61 ops/sec ±3.55%	+21.3%
priority-queue-dequeue	3,756 ops/sec ±2.42%	12,790 ops/sec ±0.27%	+240.5%

[saripovdenis]

@sindresorhus Hello! Let me know if something is unclear in this PR!