[alma-config] Conway's long message

[Mark Holdaway]

> For example,
> If the metric problem will be solved soon (say couple
> of weeks) we can postpone the simmulation until this time.

Simulations can go forward without the metrics being figured out;
we can evaluate months after the simulations.

I wouldn't worry too much about the configuration memos going into the
memo series: one reason we went over to electronic memos was so people
could just DELETE if they weren't interested.  Furthermore, I find
a high degree of interest among a wide variety of workers on the
configuration issue.

And a last comment:  it occurred to me that we don't have any experience
with the interaction of array design and a MultiScale algorithm.
This relates to my comment in the meeting: "An RMS is usually taken
at just a single resolution... or your eye is most sensitive to
artifacts at the image's resolution... and we need to evaluate the
errors on a range of size-scales."    So, I am thinking that MultiScale
algorithms may be sensitive to the details of how the samples 
fill the Fourier plane.  Or maybe not so sensitive.
I will try to investigate this, and to suggest a "multiscale image metric"
as well.

	-Mark