Empty table corners

[Antonio-Velasco]

Hi @ChrisPappalardo, I've kept using eparse, really useful :)

I've come across many instances where the tables I want to parse have an empty upper left corner to accommodate indexes and headers.
To identify these empty cells as the starting corner of the tables allows for full headers and indexes to be parsed.
I propose adding a na_depth value to the function df_find_tables that allows the user to configure how far up and right should the parser check and, if any of those cells are not empty, consider the current cell as ghost_corner rather than just empty.
With na_depth = 1 the behaviour remains the same.

PR: #7

[ChrisPappalardo]

Hi @Antonio-Velasco ! Thanks for the PR. I am at PyCon US this week and should have some time to look at this during the sprints.