make standard prior knowledge queries for bounty

[yoid2000]

I want a prior knowledge call where I can specify:

• which column I want
• which rows I want:
  • random rows
  • by fraction of all rows
  • by number of rows
  • rows matching a set of column values
  • rows falling in a range of column values
  • rows for a random set of users
  • by fraction of all users
  • by number of users

[yoid2000]

So a spec might look like this:

{
   'columns': ['col1', 'col2'], 
    'type': 'rowFrac' | 'userFrac' | 'colMatch' | 'colRange',
    'frac': 0.45,    # if rowFrac or userFrac, number between 0 and 1
    'col': 'column',        # if colMatch or colRange
    'values': [val1, val2, ...],             # if colMatch
    'range': [min, max],            # if colRange
}

[yoid2000]

completed with pull #50

[yoid2000]

done with pull #50