I'd think sekonic's warning is a little harsh - of course it CAN be measured. Whether is SHOULD be measured depends on how you interpret the results. As the results will be different it's simplest for them to tell you not to bother.

The difference is that in auto mode the flash is measuring the REFLETED light, so it puts out enough light to place your test subject at mid-grey. The sekonic meaures the light, and gives the value that would be reflected by a grey card.

If you used a perfect grey card as your test subject, the results should match up. If you used a white subject, then lots of light will be bounced back and the flash will put out less power (so the meter wil read lower).

Simlarlt if the subject is dark the flash will put out more power to get the same light bounced back.

The readings are different, because they're measuring different things. In a perfect test environment they'd probably match up.