Allow non-existent checkpoint location path in index validation

[dai-chen]

Description

The pre-validation of checkpoint locations, introduced in PR #297, requires the specified checkpoint path to exist. However, Spark is designed to automatically create necessary sub-folders when a streaming job begins. This PR is to relax this strict validation to enhance user convenience and ensure backward compatibility.

Testing

For checkpoint location without access permission, the behavior is the same as before:

spark-sql> CREATE SKIPPING INDEX ON ds_tables.http_logs 
...                (clientip VALUE_SET) 
...                WITH (
...                  auto_refresh = true,
...                  checkpoint_location = 's3://test/test'
...                );

java.lang.IllegalArgumentException: requirement failed: 
  No permission to access the checkpoint location s3://test/test

For checkpoint location with non-existent sub-folders, the validation can pass now and Spark streaming job creates it when start:

# validation-test folder doesn't exist
spark-sql> CREATE SKIPPING INDEX ON ds_tables.http_logs 
...                (clientip VALUE_SET) 
...                WITH (
...                  auto_refresh = true,
...                  checkpoint_location = 's3://daichen/validation-test'
...                );
Time taken: 11.97 seconds

# both validation-test and validation-test/subtest1 folder doesn't exist
spark-sql> CREATE SKIPPING INDEX ON ds_tables.http_logs 
...                (clientip VALUE_SET) 
...                WITH (
...                  auto_refresh = true,
...                  checkpoint_location = 's3://daichen/validation-test/subtest1'
...                );
Time taken: 12.978 seconds

Issues Resolved

#65

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[seankao-az]

Thanks for the change!