I'm experimenting with liquid clustering and have some questions about compatible types  (somewhat similar to Liquid clustering with boolean columns ).

Table created as

CREATE TABLE IF NOT EXISTS <TABLE>
(  
    _time DOUBLE
  , timestamp TIMESTAMP_NTZ
  , aid STRING
  , aip STRING
  , cid STRING
  , TargetProcessId BIGINT
)
USING delta CLUSTER BY (timestamp,aid,TargetProcessId) LOCATION <LOCATION>

Table is filled via .writeStream() so I'm under the impression that data won't be clustered on write.   Therefore, I run

OPTIMIZE <TABLE>

which results in (with "spark_version": "14.2.x-cpu-ml-scala2.12"):

IllegalArgumentException: Unsupported datatype 'TimestampNTZType'
com.databricks.backend.common.rpc.SparkDriverExceptions$SQLExecutionException: java.lang.IllegalArgumentException: Unsupported datatype 'TimestampNTZType'
	at com.databricks.sql.io.skipping.Classifier$.getStatsForCol(Classifier.scala:180)
	at com.databricks.sql.io.skipping.SimpleClassifier.applyActionToIntersectingBinaryNode(Classifier.scala:424)

I was able to create the table with the CLUSTER BY command so I assumed the column list was acceptable,  but this exception indicates otherwise.   If TimestampNTZType is an unsupported datatype,  where is this documented?

When I run DESCRIBE on the table, it indicates the cluster columns were accepted: