Just to clarify: My remarks, now and before, are mostly valid for the "small islands" setting. Depending on the exact archipelago map script you may or may not have enough land to have basically a land-based economy with lots of water tiles, or are forced to mostly use coast tiles.
Of course coast tiles are really bad, no matter what, so if you can use land, how you can use the little land you have will be most important. But in small islands, most of your cities use mostly coast tiles and any UB that improves those coast tiles becomes good. So I'd say that, yes, the Portuguese UB is one of the worst UBs in the game (a base building you don't typically want to build, which now also changes a mediocre/bad tile into a mediocre tile, but you still usually don't want to work coasts even with a Feitora), but if most tiles you work are coast anyway, then it's better than a lot of other UBs. The Dutch UB still comes rather late in the game, but I find small islands maps play slower than other maps (slower expansion, less early wars), so that is less of a problem on these maps, and because you work lots of coast tiles anyway, it increases output significantly on a production-starved map (and have you ever had a golden age on a Dike-filled small-islands map).
Also, it should be noted that with small islands you can usually get trade routes with lots of far away people as those coastal areas often interconnect. Often you can trade with more people pre-Astro as on a continents map. So the international trade benefit of Astronomy is not as big as it might seem. Typically you do have intercontinental trade routes within your empire, so it is only the "foreign" bonus that you miss. Thus you have to work only roughly 1.5 (if there are no big foreign cities) coast tile per new international trade route to make Colossus come out on top. I generally lose commerce when researching Astronomy.