Investigate and improve allocation behavior of collapse-perf

Since the project is fairly recent I assume it's benched on a recent version of Rust (or even nightly, does Criterion require nightly?) That means it's using the system allocator, and IIRC flamegraph munging is a lot of string munging and other allocation-heavy tasks, jemalloc could well have an edge.

Start with the collapse-perf entry point and its Criterion benchmark setup, then compare allocation behavior under the system allocator and jemalloc. Use the flamegraph munging path as the focus and treat measured allocation or performance improvements as the completion criteria.

Written by the indexing model from the issue text.