Presto not showing microseconds for table created via spark-sql

[tooptoop4]

Use spark-submit (v2.4.6) to write parquet to s3://redact/fol/part_filename=ju.txt/ (parquet is https://github.com/tooptoop4/presto-1/blob/master/part-00000-e35ff4f1-5a4b-4b5e-afde-fe00fcc1a7b2-c000.snappy.parquet?raw=true)

Use spark-sql to run:
CREATE EXTERNAL TABLE s.t (opr_date TIMESTAMP,city STRING,hour DOUBLE,temp DOUBLE,temp_diff DOUBLE,temp_normal DOUBLE,dew DOUBLE,cloud_cover DOUBLE,feels_like DOUBLE,feels_like_diff DOUBLE,precip DOUBLE,wind_dir DOUBLE,wind_speed DOUBLE,hdd DOUBLE,cdd DOUBLE ) PARTITIONED BY (part_filename STRING) STORED AS PARQUET LOCATION 's3a://redact/fol/';

Use spark-sql to run:
MSCK REPAIR TABLE s.t;

Query in HiveServer2 (v2.3.4):
select * from s.t;

Query in PrestoSQL (336 and JDBC driver is 336):
select *,date_format(opr_date,'%f') from s.t;

show create table s.t;

CREATE TABLE s.t (
opr_date timestamp(3),
city varchar,
hour double,
temp double,
temp_diff double,
temp_normal double,
dew double,
cloud_cover double,
feels_like double,
feels_like_diff double,
precip double,
wind_dir double,
wind_speed double,
hdd double,
cdd double,
part_filename varchar
)
WITH (
external_location = 's3a://redact/fol',
format = 'PARQUET',
partitioned_by = ARRAY['part_filename']
)

cc @martint

[findepi]

This is covered by #3977