remove inline declarations and refactoring

[cxxxr]

I removed the inline declaration and refactored the code. I couldn't think of any particular advantages to this inline declaration, but as a disadvantage of inline declarations, debugging information is lost and the code becomes less readable. I also think it is better to optimize the internal implementation rather than inlining the exported functions.

[cxxxr]

A benchmark has been added. Comparing before and after refactoring, there was no effect from the inline declaration.

before

Evaluation took:
  2.750 seconds of real time
  2.741348 seconds of total run time (2.681467 user, 0.059881 system)
  [ Real times consist of 0.154 seconds GC time, and 2.596 seconds non-GC time. ]
  [ Run times consist of 0.152 seconds GC time, and 2.590 seconds non-GC time. ]
  99.67% CPU
  9 lambdas converted
  6,019,819,039 processor cycles
  5,917,078,928 bytes consed

after

Evaluation took:
  2.775 seconds of real time
  2.764296 seconds of total run time (2.702289 user, 0.062007 system)
  [ Real times consist of 0.150 seconds GC time, and 2.625 seconds non-GC time. ]
  [ Run times consist of 0.153 seconds GC time, and 2.612 seconds non-GC time. ]
  99.60% CPU
  9 lambdas converted
  6,074,091,252 processor cycles
  5,917,121,456 bytes consed

[fukamachi]

Do you have any actual problem with it? I'm curious.

[cxxxr]

Looking at examples of its use in jsonrpc, it is referenced from multiple threads, so it is necessary to make it thread-safe. This is a preliminary refactoring for that purpose.

[fukamachi]

Fine, as I don't have a chance to use this library anymore. I suppose it was not for anything about performance since event-emitter is quite slow and does not care about speed.