True enough, the building code (or, more commonly the NEC) generally applies
to permanent stuff (although "cord connected" stuff gets covered, as do
certain temporary installations).
The safety grounding aspects are more the province of "good engineering
practice" and the requirements imposed by testing labs such as UL. There's
also good grounding info in a few IEEE specs.
Your distinction is a good one. If you are indoors, the "ground" that a
person is connected to (by capacitance, if nothing else) is a bit trickier
to figure out than standing out in your driveway.
In any case, you want the RF currents to flow where YOU want them, not
somewhere else, and particularly not through the building wiring system.
However it's also important to think about what fraction of the RF current
is carried by the streamer (a small amount, I'd venture). Most of the RF
current (which is what will cause the problems) is just capacitively coupled
from topload to the other end of the secondary.
If you have some wiring in the field of the TC (and there is), you'll get RF
coupled to it. Much better to have the return path directly to the coil by
bonding your counterpoise locally, rather than routing the RF through the
wiring to the panel ground, then to your ground rod, then back up to your
coil.
> If OUTSIDE, your counterpoise is earth and the TC should be directly
> grounded to it. If all objects and persons are at a safe distance, > there
> is no fault that can raise anything outside this safe distance to a
> hazardous voltage.
As long as you can maintain that "safe distance" you're good to go. This is
more challenging in practical terms than one might think.