I got a response back that confirmed that this machine only is supported
with CS/MS.  I would not expect S3 to work properly.

Thanks for the confirmation, that's what I expected following the feedback on
other recent Dell machines.

Even if S3 is part of the ACPI namespace, the behavior that Windows follows
is that S3 is not offered when the low power idle bit is set in FADT.

I know that the current behavior in Linux does not review low power idle
bit to determine whether to offer S2I or S3

Is it something that is envisioned already as a future evolution in Linux as
well, if more and more systems are now designed with S2I in mind?

so I'm inquiring if we can remove
S3 from a future BIOS update on this platform to let Linux work better
even though we don't ship this platform with Linux.

That would be a great fix, removing the default to S3 for this system on Linux.

For now I would recommend that you focus efforts exclusively on making
S2I work, and ignore S3 problems.

Well noted!

It's unclear to me exactly when you tested s2idle-dell-test.
Can you re-test with Rafael's newest submissions and changes to s2idle-dell-test
committed 28 hours ago?
You'll still have to add your system to the DMI whitelist for the last patch
but he has reworked a bunch of related wakeup code that has also changed.

I've just done so, using the updated s2idle-dell-test branch + my DMI
adaptation.

I still get exactly the same results than during my previous tests:
- it provides the expected fix when running BIOS 1.1.20: short press to wake-up
- it doesn't fix with the current BIOS 1.1.31: long press still needed
to wake-up
so something has clearly changed between these 2 versions. Btw, the EC is
explicitly updated when flashing version 1.1.31, that's one of the
visible steps.

Some assumptions now: either the SCI is ignored erroneously or it's
not interpreted correctly by the expected driver? I guess that's the 2
possibilities I'll try to investigate.