Add validate method to synthesizer

[amontanez24]

Problem Description

As a user, it would be helpful if I could check if my data was valid according to my metadata.

Expected behavior

• Add a validate method to the BaseSynthesizer
• If the method finds an errors, it should raise an InvalidDataError in the following format

synthesizer.validate(data)
InvalidDataError: The provided data does not match the metadata

Error: Invalid values found for numerical column 'age': ('a', 'b', 'c', +more)

• The method should do the following checks
  • Columns described as sdtype=numerical should only have data that is numerical (or missing) – no other strings, etc. are allowed
    Error: Invalid values found for numerical column 'age': ('a', 'b', 'c', +more)
  • Columns described as sdtype=datetime should only have data that datetime or missing – try to convert everything to the datetime format and error if it fails
    Error: Invalid values found for datetime column 'start_date': (0.0, 30.0, 4.4, +more)
  • Columns described as sdtype=boolean should only have values that are: True, False or missing
    Error: Invalid values found for boolean column 'is_subscribed': (0.0, 30.0, 4.4, +more)
  • Columns marked as a key (primary, alternate, sequence, or foreign) should not have any missing values
    Error: Key column 'user_id' contains missing values
  • Columns marked as a primary or alternate keys should be unique in the table
    Error: Primary key column 'user_id' contains repeating values: ('UID_000', 'UID_001', 'UID_002', +more)
  • (Sequential only) Context columns (stored in the model's parameters) should be fixed for each sequence – ie if you group by sequence key, the context columns should not vary
    Error: Context column 'patient_address' is changing inside sequence ('Patient_ID'='ID_004').

Additional context

• For the sequential case, we should override the method in the PARSynthesizer