You say you have 18,000 data files.  are those all on different disks?  If not, you will not really get a performance benefit and I can see that causing a...

I might have an idea - how many VLF's do you have?
DBCC LogInfo(<database name>)

If you have a crazy high number of VLF's, that will greatly...

Did you look in the event log?  I am willing to bet that that is exactly what it shows in the event log.  I've sen those types of messages in...

I've not done much (ie anything) with windows failover clustering,but that dependencies list doesn't look like it lists cluster disk 5 as being included in the failover.

As a...

9 TB of data is a lot of data to restore, so it doesn't surprise me it takes a while especially on a system with that low of memory.

Right.  About the CHECKDB thing.  that threw me off.  But you have not run checkdb in quite some time, no?   2017-01-19 appears to be the last time it was...

If I understand the logic correctly, I think this is more what you want for your loop:
WHILE EXISTS(SELECT 1
             ...

Well, first I would add in the RECOVERY option to the restore.

Second, it sounds like you are out of resources to do the restore.  how much free space...

I think to encapsulate it you would need to put it into a UDF but as you said that'd be a RBAR thing and you are trying to avoid those....

The other option, presuming you have SQL Enterprise or SQL 2016 SP1, is to use change data capture.
https://technet.microsoft.com/en-us/library/bb522489(v=sql.105).aspx

how about this:
SELECT CAST(ABS(CAST(CAST(FirstDate AS DATETIME) AS INT) - CAST(CAST(SecondDate AS DATETIME) AS INT)) AS INT)/365
Not positive that outer cast to an INT is required, but...

I did this recently on our test box.  I just used the SQL 2016 installer to do this.  I clicked on the upgrade button and let it do its magic....