Handle invalid Transport and fix problem on async loop during error

[TomChv]

I used winston-elasticseach to send logs on my ELK. I had some problems with auth on ElasticSearch and that made my server run really slowly because winston couldn't send logs....
I made this pull request to fix those 2 problems:

• Handle invalid transport: It will not emit an error event then, you can handle it as you want in your codebase.
• Handle error on failed report: It will emit an error if one of the bulk reach the max attempt. I think there is an other way to handle that but I give you a begin of idea.

[TomChv]

This PR can also fix #115 Issue, I'm waiting for your comment to clean the code.
I can also refacto the bulkWriter into a class if you want.