Paul,

You make some interesting points.

The real question is the grounding impedance.
measuring the grounding impedance at a few hundred kHz
might be a bit problematic.

Most sources claim the energy is concentrated between DC and a few tens
of kHz. This means that impedance is an issue, but not a terrible one.
Typical faraday cages (few make lightning "rods" anymore...) are
anything but low impedance, yet thick enough to take 100kA without
melting.

Assuming the direct lightning hit is about 10-30 kA,
a grounding impedance of only 1 ohm would create
a potential difference of 10-30 kV.

Right. Perhaps even worse on the high end of the spectrum.

the grounding impedance should be measured
between the antenna ground and house ground.

This reminds me of a web page from your corner of Europe:
http://www.kolumbus.fi/oh5iy/back/Ham%20Radio.html

The take-home point is that one wants all sensitive equipment (and
people) to be on a high-insulation, high-impedance branch of a
multipath graph, while bolt energy is offered a low-insulation,
low-impedance branch to run through.

There are interesting implications:
Assuming that the house has a separate grounding electrode [...]
due to the finite ground resistance, the house grounding electrode
will move to an elevated potential (several kV) compared to the surrounding
mains neutral, telephone and CATV).

Not necessarily. If grounds are separate, inter-groundpoint resistance
will if anything decrease the risk of such an transient. If there was
a single grounding point, and ground resistance were high, there would
be a greater risk of the grounding point becoming a source of common
mode transient towards the land lines.

In my case, my shack is at the ground floor of a 9-story building, the
antenna is on top, and all grounds are pretty heavy duty and bonded -
but not low impedance. The risk for the land lines would be about the
same w/ or w/o my puny 20m-long T2FD. (The Other Guy has a 35ft self
supporting lattice with tribander, 4x15 @ 144, 3x32 @ 430, 2m dish @
1296, weather turnstyle, 40/80 dipoles, collinear @ 144, and a 6m
moonbounce dish @ 430. When I added the T2FD nobody noticed).

As for the DC block, I think I found the right container and location.
A 1 - meter PVC pipe that will hang horizontally under a roof edge less
than 2m from the flue lines. Pipe and coax can tiptoe away from the
chimney in a environment that stays dry when it rains.