Define DataFrame operations where Schema will always remain valid

- There are some DataFrame operations that will always keep the Schema valid (`sample`, `head`, `copy`, etc) where it could be beneficial to skip the `_get_invalid_schema_message` check.

If we want to save the time spent checking whether the schema is still valid, we can define a set of DataFrame operations that cannot invalidate the schema and, if the `attr` supplied to the Accessor falls into that set, we skip the validation check.