I have a Personal cluster version 15.4 LTS (includes Apache Spark 3.5.0, Scala 2.12) and a SQL Warehouse in a databricks environment.  When I use the following code to create a table in a catalog, it gives me different column types when run on the cluster vs the warehouse:

%sql
create or replace table [catalog].[schema].[test_table_name] 
using delta
comment 'This has a comment'
as
select 
  id, 
  name as new_name,
  created_date as new_created_date,
  current_timestamp() as test_timestamp,
  coalesce(name,'Replaced name') as test_coalesce_name,
  coalesce(id,'-1') as test_coalesce_id,
  coalesce(created_date,'2024-12-11') as test_coalesce_date
from (
  select 
    cast(col1 as int) as id, 
    cast(col2 as string) as name, 
    cast(col3 as date) as created_date
  from VALUES
  (1, 'Alice', '2024-12-01'),
  (2, 'Bob', '2024-12-02'),
  (3, 'Charlie', '2024-12-03'),
  (4, 'David', '2024-12-04'),
  (5, 'Eve', '2024-12-05'),
  (6, 'Frank', '2024-12-06'),
  (7, 'Grace', '2024-12-07'),
  (8, 'Hank', '2024-12-08'),
  (9, 'Ivy', '2024-12-09'),
  (10, 'Jack', '2024-12-10'),
  (11, NULL, '2024-12-11'),
  (NULL, 'NULL Values', NULL)
) as temp_table

When running on a SQL warehouse, the column types for the coalesce'd columns are resolved correctly. However, when running on a cluster, they are not resolved and are converted to strings.  Is this expected behaviour?

Have tried on two different databricks environments and have the same result.