Wrong documentation for validate_instance()

[kennysong]

In the TFDV Get Started page, it states that:

TFDV also provides the validate_instance function for identifying whether an individual example exhibits anomalies when matched against a schema. To use this function, the example must be a dict mapping feature names to numpy arrays of feature values.

However, validate_instance() does not work on dictionaries. It only works on Arrow RecordBatches.