Currently schema validation only throws expected and actual schemas.
For simpler datasets it's pretty easy to figured out what's wrong but for schemas with 20+ columns a simple typo might take forever to detect.

Instead of printing expected, actual we should print:

• mismatched entries - when entry name is the same but type is different
• missing entries - when entry is defined in expected schema but not present in given
• extra entries - when entry is not defined in expected schema but present in given

Besides we should make sure that when expected schema is nullable and given is not, it wont trigger schema mismatch exception