Yes, it could be scientifically detectable. If we could make predictions about what that would mean. In archeology, intelligence is obviously detectable as we can predict that, for example, intelligence implies use of tools, making new tools and maybe building houses. Then, if we dig and find the foundation of a house, some tools and ceramics and so on, we have evidence that this wasn't just nature, but something/someone intelligent. If we only found a rock that looked like a tool every now and then, no more than statistics might predict, then intelligence might be disproven. This same method applies to all the fields of science you mentioned. I think that any real scientist does not reject it out of hand, but rather because assuming intelligent design doesn't allow us to make useful predictions about fossils we should(n't) find.

As for the last point, about there being no unevolved intelligence, as far as I understand, that is perfectly reasonable. Unless intelligence was built-in from the start (single-celled life) it first had to evolve, like ANY other feature of life as we know it. The only other possibility that I can see would be extra-terrestrial intelligence, which would have evolved somewhere else and would to us seem unevolved, appearing suddenly. That is, however, irrelevant until you can show me evidence that intelligence was involved at all!

"Any intelligence responsible for biological complexity would have to be an unevolved intelligence"

I don't understand why this would have to be so? Human's, as an evolved intelligence, is able to understand most of the complexities of the human, and other species' body, so we could, I would suggest be able to design a simple multi-cellular organism.

Any who, back to detecting this "intelligent design". SETI looks for extra-terrestrial intelligence by looking for repeating EM radiation in the regular EM background radiation. If you where to adopt a similar approach to ID, I would suggest you would have to find a common marker, a signature between all life. Slight variations in cell design and function would suggest evolution. Exact replications of cells in all life could be construed as ID.

It is up to the ID crowd to suggest plausible criteria for science to look for so that it can be tested. So far as I judge it, they instead choose to misrepresent existing data or research, pretending that it supports their case regardless of merit, or go outside the peer review process to avoid criticism of their "findings." It's not surprising, since their goal is to provide evidence supporting a preformed assumption, not to test a theory. If ID is provable by scientific methods, the onus is on ID proponents to do so in peer-reviewed legitimate forums, or at worst present legitimate avenues of investigation, not on scientists to bend over backwards to do the hard work for them.

In order to 'test' for intelligent design (ID), first some criteria would have to be agreed upon about what features one could expect to discover through testing if intelligent design was present.

This creates a problem in actually designing a test, because it is precisely these criteria which is central to the disagreement between conventional science and proponents of ID.

As one example, someone arguing for ID may state that certain systems in nature are 'irreducably complex' (that they are extremely unlikely to have come about through random mutation). Setting aside the details argument, for now, the vast majority of scientific experts disagree that irreducable complexity is an issue. In other words, evolutionary scientists and ID proponents disagree on the basic criteria with which we might be able to detect ID.

This leaves a dilemma: if, in testing for ID, we use criteria devised by ID supporters, the result will almost certainly be positive ('Oh, look! An eye - that's irreducably complex! Test positive!') but will not be accepted by the majority of scientists ('Actually, it's a bit more complicated than that...'). If we use criteria proposed by people in agreement with generally accepted science, then - assuming the test is negative, as is likely - the ID supporters will likely not be satisfied, denying that this has answered any of their stated problems with evolution.

The only way out of this dilemma is if we use the scientific criteria, and the tests prove positive for ID - because that is the only outcome in which both parties would agree that it was a fair test! Since evolution without the need for ID is such a well supported theory, though, this outcome seems unlikely. Thus, even though it is conceptually possible to 'detect' ID, it looks bleak for devising a test that people on both sides of the argument will agree on.