Note that it is customary to restrict posts to a single question, which is easier to answer for these complex topics. I'll do what I can here. I'm going to deal with everything at once and then reference pieces in direct reference to your 3 questions.
MECHANISMS OF CONSTRAINTS ON PLOIDY
I think that it's worth separating out some of the different mechanistic ways in which a practical limit can be set for how many chromosome copies you could have.
Another point is that the examples you give in flies and silkworms are organs with high ploidy, rather than whole organisms. It's a lot easier for an organism to have a few weird snippets of tissue with weird ploidy than to maintain that ploidy in the germ line.
Ciliates are weird and instructive:
As another totally insane example of how some organisms have evolved to reconcile this somatic desirability of lots of DNA with a stable germ line, you should look into the nuclear organization of the ciliates, which have a germ-line micronucleus (which is just passed on to future generations) and a somatic macronucleus (which is generated by amplification of the micronucleus and does all the cellular work of DNA). The macronucleus is of extremely high "ploidy" (it's more complicated than that but that's close enough), but the germ line is of a more sensible ploidy.
Some constraints I could think of:
Physical limits:
DNAs are basically huge polymers. Packing polymers into a cell imposes a physical constraint, in the sense that you need to 1) fit all that polymer into the cell/nucleus, and 2) need to still be able to use the DNA for all of the necessary cellular processes. These constraints will vary vastly from organism to organism, based on cell volume, plasticity, morphology, etc.
Metabolic limits:
DNA is expensive to make, energetically and physiologically. It ties up a lot of sugar and nitrogen that you could use to do other more interesting things, like the rest of cellular metabolism. Silkworm silk glands are obviously eye-popping, but I assume that there is some theoretical limit above that million-ploid amount (for silkworms) that would be too much DNA to support. For a little more information see this Mike Lynch paper: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4697398/
Chromosome segregation limits:
Moving homologous chromosomes to daughter cells is not a perfect process; cells mess it up frequently, leading to aneuploidies and cancers. Some ploidies such as triploid, and presumably higher numbers with 3 as a smallest common denominator, are particularly unstable in meiosis. However, any high ploidy leads to errors at a higher rate than lower ploidies. Look at the ciliate example also. This is I believe the main reason for the "collapse of high ploidies into lower ploidies that you mention.
UPPER LIMIT TO PLOIDY IN NATURE
See all above answers.
UPPER LIMIT TO SYNTHETIC PLOIDY
See all above answers. Synthetic increases should behave the same as natural ones, as we are honestly much worse at getting making things happen than nature itself. Nature will have gotten higher already than we ever will, in all probability.
PURPOSE OF HIGH PLOIDY
See the cost/benefit of ploidy article from above.