Adds support for config.properties file to adjust settings for CoreNLP

[raphaelhoffmann]

One can now add a config.properties file into the directory where one is calling Pipe. These settings are then passed through to the CoreNLP modules that are called. For example, if you want to parse pre-tokenized and pre-sentence-split content, you can add a config.properties file with the following entries:

tokenize.whitespace = true
ssplit.eolonly = true

[netj]

What about Parser? Does the config affect Pipe only?

[raphaelhoffmann]

Adds support for config.properties to Pipe. @netj Not sure if anyone is still using the old Parser code, maybe better to converge on one project to avoid duplicate effort.