### What changes were proposed in this pull request?
This change is to support user provided nullable Avro schema for data with non-nullable catalyst schema in Avro writing.
Without this change, when users try to use a nullable Avro schema to write data with a non-nullable catalyst schema, it will throw an `IncompatibleSchemaException` with a message like `Cannot convert Catalyst type StringType to Avro type ["null","string"]`. With this change it will assume that the data is non-nullable, log a warning message for the nullability difference and serialize the data to Avro format with the nullable Avro schema provided.
### Why are the changes needed?
This change is needed because sometimes our users do not have full control over the nullability of the Avro schemas they use, and this change provides them with the flexibility.
### Does this PR introduce _any_ user-facing change?
Yes. Users are allowed to use nullable Avro schemas for data with non-nullable catalyst schemas in Avro writing after the change.
### How was this patch tested?
Added unit tests.
Closes#30224 from bozhang2820/avro-nullable.
Authored-by: Bo Zhang <bo.zhang@databricks.com>
Signed-off-by: Gengliang Wang <gengliang.wang@databricks.com>
551b504cfe