fix: Schema validations ignore not null on Teradata and BigQuery

[nj1973]

This change adds code to correctly identify nullable True/False on Teradata and BigQuery, bringing them into line with other engines (MySQL, Oracle, PostgreSQL, Snowflake, SQL Server).

It also adds tests for all of these engines to avoid regressions.

It does not add support or tests for these engines:

• Hive: Raised issue schema validation: Nullable=False not identified on Hive #934
• Spanner: The Spanner client does not include nullable information, it appears to get data types from a cursor of a SELECT * ... LIMIT 0 query which does not expose schema attributes. I feel this is too far removed from DVT to tackle

We did talk about adding a flag to give users the option to ignore nullable True/False for schema validations, this would be useful for Hive and Spanner. But I decided that it would be a feature request and not a function of this issue.

[nj1973]

Please do not review this yet. I am getting a Python Failed to build thrift error which I do not understand because I haven't added any new packages. What I have done is import bigquery in a different file so perhaps that has had a knock on effect.

[nehanene15]

Please do not review this yet. I am getting a Python Failed to build thrift error which I do not understand because I haven't added any new packages. What I have done is import bigquery in a different file so perhaps that has had a knock on effect.

You may want to try re-running the build to make sure it's not an ephemeral error

[nj1973]

/gcbrun

[nj1973]

You may want to try re-running the build to make sure it's not an ephemeral error

@nehanene15 , you were right, the build failure was transient. Once the test had run they highlighted other issues for me to work on.

This should now be ready for review.

[nehanene15]

LGTM!