Use jemallocator

[Razican]

I have changed the system allocator for Jemallocator as the global allocator for the binary and the benchmarks. Jemalloc makes the binary bigger, but the performance should increase nicely, which is something we probably want in a compiler.

My benchmarks show around 10% improvement or more in some cases.

[github-actions]

Benchmark for f1943c6

Click to view benchmark

Test	PR Benchmark	Master Benchmark	%
Create Realm	360.5±15.93µs	409.0±21.64µs	87.00000000000001%
Expression (Parser)	2.8±0.27µs	3.1±0.09µs	87.99999999999999%
Hello World (Execution)	357.1±20.79µs	369.5±27.44µs	97%
Hello World (Lexer)	953.5±47.84ns	869.3±53.72ns	110.00000000000001%
Hello World (Parser)	1034.5±48.33ns	912.0±83.22ns	112.99999999999999%
Symbol Creation	377.8±22.31µs	466.8±54.15µs	76%
fibonacci (Execution)	4.0±0.14ms	4.0±0.20ms	100%
	undefined	undefined	100%

[github-actions]

Benchmark for 9679a78

Click to view benchmark

Test	PR Benchmark	Master Benchmark	%
Create Realm	419.8±12.13µs	470.5±13.15µs	87.99999999999999%
Expression (Parser)	3.1±0.07µs	3.5±0.09µs	87.00000000000001%
Hello World (Execution)	421.6±9.70µs	486.1±12.02µs	85.00000000000001%
Hello World (Lexer)	1186.2±51.31ns	1051.6±23.12ns	112.99999999999999%
Hello World (Parser)	1323.6±48.71ns	1333.0±37.41ns	99%
Symbol Creation	465.5±14.15µs	598.9±23.81µs	71%
fibonacci (Execution)	4.8±0.17ms	5.5±0.08ms	85.00000000000001%
	undefined	undefined	100%

[jasonwilliams]

What’s the reason Jemallocator makes the binaries bigger?
(I know very little about jemalloc)

Edit: reading https://internals.rust-lang.org/t/jemalloc-was-just-removed-from-the-standard-library/8759

[jasonwilliams]

Why are the benchmarks stuck hmm

[Razican]

I cancelled, but it seems I cannot run it again. Still, it seems it passed two times, right?

[Razican]

Should we merge this?