Optimize release and benchmark builds

[Sharktheone]

I've added some options in Cargo.toml to optimize the performance of release and benchmark builds.

• lto or link time optimizations, allows optimizations across codegen units and even languages (for example when we are using v8). It also inlines most of the functions. See this and this
• codegen-units Disallows the compiler to split code in multiple codegen-units, which results in a bit better performence.
• opt-level Level at which the compiler optimizes things. (e.g unrolling loops, ...)

This of course causes the build to take much longer. From my measurements a full release build takes about 23.3s instead of 10.8s on my machine. A rebuild (with some dependencies already built) takes 17.3s instead of 2.8s.

Development builds are untouched, so they only take 6.2s with or without these options.

The benchmark results go up ~10-11%.

Benchmark results

TreeIterator - wikipedia

TreeIterator - stackoverflow

TreeConstructor

Tokenizer

[jaytaph]

This sounds good. We dont need to worry much about prod builds, but its good to have it already

[jaytaph]

pr author check fail is fine. This probably needs finetuning or we might need to get rid of it if it doesn't work out well enough..

[Sharktheone]

pr author check fail is fine. This probably needs finetuning or we might need to get rid of it if it doesn't work out well enough..

Probably because I didn't add myself to the file.

[jaytaph]

maybe, but that check is still broken :-)