Because the code is different :)

True, the code is different, but the expressions are all equivalent: they're all 'degenerate' single-item slices.

Because @warnme... is different from @$derefNowarn...

I don't understand the significance of this statement. Are scalar derefs in this context (single-item slices) simply not warned about? If so, is there a rationale for this? (I could understand if this was simply beyond the scope of the warnings system as it now stands. After all, you can't warn about everything.)

For verbosity see diagnostics...

diagnostics are indeed more verbose, but don't throw more light, for me, on this (admittedly somewhat trivial) problem. Oh well, going to bed now...