Switch to DataFrames and add Metrics Filters

[kylemann16]

This pull request makes a couple of major changes:

1. Convert from using mostly numpy ndarrays to pandas DataFrames

   • Dataframes are easier to work with, and easier to manipulate overall
   • Dataframes make it easier to implement the filters feature, as it's easier to pass around DataFrames with select attributes

2. Add Filters to Metrics

   • Filters will act as a list of functions that manipulate a DataFrame before it arrives at the Metric's main method.
   • This will allow for users to add thresholds, like an elevation threshold for a specific metric

3. Split up methods a bit more in Shatter so that computationally heavy pieces have more room to operate.

4. Add Attribute dependencies for a Metric so we can be sure that the database and any ingested data has that attribute

5. Change Extract to be using pandas DataFrames as well.

   • Added better filtering and overlap detection