Apparently I keep forgetting to bring this up, but I see again that
we do not seem to have any accepted method for storing the TR of a
timeseries dataset.

Did we discuss this? Maybe a timestep MetaData element at the GIFTI
level would work for that. We are applying it as AFNI_timestep here,
but of course everyone will need that piece of information.

I see that our MetaData elements are capitalized, so how about using
Timestep or TimeStep as the MD element at the GIFTI level?

- rick