fix: Avoid changing table configs when creating a table with an existing base path on Spark

[yihua]

Describe the issue this Pull Request addresses

When creating an external table using the following CREATE TABLE SQL statement on Spark, the table config file .hoodie/hoodie.properties is updated with additional table configs, hoodie.table.create.schema and hoodie.database.name. However, these two table configs are not needed for an external table, especially for reading the table only. Updating the table config requires write access which is not needed for read-only case, causing confusion to users.

CREATE TABLE IF NOT EXISTS table_name USING hudi LOCATION '<table_base_path>'

Summary and Changelog

This PR fixes the issue above, to keep the table config the same during executing CREATE TABLE SQL statement if the table's base path exists, where the change to table configs is unnecessary.

A new test is added to validate the change.

Impact

Improves CREATE TABLE SQL experience.

Risk Level

low

Documentation Update

N/A

Contributor's checklist

• Read through contributor's guide
• Enough context is provided in the sections above
• Adequate tests were added if applicable

[hudi-bot]

CI report:

• 1752db8 Azure: SUCCESS

Bot commands

@hudi-bot supports the following commands:

• @hudi-bot run azure re-run the last Azure build