Create performance impact tests/reports

[lance]

Code tracing can be expensive. How expensive? Well, we don't know yet. We should produce reports with the build that show what the cost of enabling tracing is.

[helio-frota]

I dunno if genet can help us, since genet does not accumulates calls :/

[helio-frota]

• I dunno if is correct to sum all genet lightbright related calls
• If yes ^ maybe we can create a filter or something to separate only lightbright related calls, then we can see some impact.

[1ms]-------------[2ms]------------[1ms]-------------[1ms]
functionA() --> lightbright() --> functionB() --> lightbright()

lightbright() : 3ms

makes sense ?

[helio-frota]

@lance https://github.com/panther-js/lbinfo#lbinfo

[helio-frota]

I think travis is not the 'correct' place to do a bench test
This is the 'more correct' results from lightbright:

hf@archT440 ~/p/lightbright/benchmark ±lightbright-4⚡ » node benchmark2.js
With lightbright x 99,825 ops/sec ±0.62% (92 runs sampled)
hf@archT440 ~/p/lightbright/benchmark ±lightbright-4⚡ » node benchmark3.js
Without lightbright x 102,732 ops/sec ±0.89% (86 runs sampled)
99k ops/sec with LB and 102k ops/sec without LB

Running without LB first on CI, will consume resources and causing this:
https://travis-ci.org/bucharest-gold/lightbright/jobs/172602367#L612
https://travis-ci.org/bucharest-gold/lightbright/jobs/172602367#L621

[helio-frota]

bucharest-gold/entente#133