Use jemalloc as default allocator.

[nihohit]

This brings a significant perf improvement in Rust & Node. This replaces the allocator only for GLIDE in the Node library, and replaces the allocator for the whole Rust benchmarking application.
jemalloc.xlsx
Python jemalloc results are for code that isn't added here, where I used jemalloc in a similar way to the Node system. Since there wasn't an improvement, that change isn't a part of this PR.

Issue #, if available:

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[ikolomi]

Generally its somewhat of a dirty practice to super-optimize the bench-marking tools instead of making them as simple (or rely on standard components) as possible yet able to utilize capabilities of SUT. I am ok to switch to jmalloc at our side if a. Its stable b. have a tangible benefits.
What are the results when only the node binding is changed to jmalloc?
Why would we try jmalloc for the core/redis-rs as well?

[shachlanAmazon]

What are the results when only the node binding is changed to jmalloc?

Those are the node results in the file. The Rust benchmarks tests only Rust code. The node benchmark (which isn't changed here) is written in TypeScript.

Why would we try jmalloc for the core/redis-rs as well?

This is the change - the core/redis-rs code + the Rust bindings are changed to use jemalloc.

[shachlanAmazon]

on a local run of npm run build:release, this adds 700K to the node binary size - from 4.7MB to 5.4MB.