What is going on with the SQL Server query optimizer/planner?

filtered-index-weirdness.sql

drop table if exists [dbo].[Simple]
go

CREATE TABLE [dbo].[Simple](
	[Id] [int] IDENTITY(1,1) NOT NULL,
	[Name] [nvarchar](50) NOT NULL,
	[Extra] [bit] NULL,
	[Data] [nvarchar](max) NULL,
 CONSTRAINT [PK_Simple] PRIMARY KEY CLUSTERED (	[Id] ASC )
)
GO

-- Data has potentially huge amounts of data, thus not included
CREATE NONCLUSTERED INDEX [IX_HasData] ON [dbo].[Simple]([Name] ASC) WHERE ([Data] IS NOT NULL)

go

set NOCOUNT ON
set STATISTICS TIME OFF
insert into Simple(Name) values ('Name')
go 500000

set NOCOUNT OFF
set STATISTICS TIME ON
update Simple
set Data = 'blah blah blah blah'
where Id % 100 = 0

update Simple
set Name = Id % 29

go

-- check query plans
-- why is IX_HasData not used in the first 2 cases? The last 2 cases emulates what I expect the query engine to do or a key lookup
select * from Simple where Data is not null
select * from Simple where Data is not null and Name = '25'
select * from Simple where Id in (select Id from Simple where Data is not null)
select * from Simple where Id in (select Id from Simple where Data is not null and Name = '25')

-- why no key lookup?
select Id, Name, Extra from Simple where Data is not null and Name = '25'

-- this is ok
select Id, Name from Simple where Data is not null and Name = '25'

-- what it should be doing
select * from Simple with (index = IX_HasData) where Data is not null and Name = '25' OPTION (MAXDOP 1)