Recover from tests that loop

Here's an idea:

The `lspec` binary would search for a text file containing some configurations for LSpec, including the maximum time allowed for each test file to run. The challenge, then, is to kill the task programmatically if the consumed time is exceeded.

Extra: eventually the config file could contain more customizable options like tests to be skipped etc.