Support a wider variety of validation options

[xaviernogueira]

We may want different options here.

• Lazy (based on dtype).
• Exhaustive (checks every row, currently this is what is implemented).
• First N rows check.
• 1-to-1 (columns and the model must match exactly).
• Inclusive (there can be many extra columns as long as the model schema is met, this is what is implemented).

I am going to think a bit more on how to have these options without bloating up functions.