hi, I've just come across this looking for an answer to a related issue. You may have already found your answers, 1 year later, but in case you're still curious:
the small hairlines around the flattened watermark are what Adobe calls "atomic regions." For transparent type, or any vector shape with a transparency effect (e.g. drop shadow) to be flattened into the artwork behind it, it needs to be rastered into a bitmap to burn it into a single object. The atomic regions are small slices of image where that vector-bitmap rastering happened. In virtually all cases you will not see that in print. (If you zoom in and it gets larger, then you may. It typically stays the same size or disappears.)
There's a host of settings in both versions of Acrobat (more in 9) wherein you can determine how much "extra" information gets carried into your file. Even after flattening within Acrobat, you can still optimize the file size by tossing all of that extra info. (go to Help menu and enter Reduce File Size)
What I suspect is happening is, when outputting a flattened X-1a from Photoshop, it's taking care of rastering the watermark into the image, creating a flattened file. (this is assuming you're not applying the watermark in Acrobat.) This is a single bitmap image.
If you're flattening say, an X-4 (unflattened) PDF with live transparency, or burning in an Acrobat watermark, Acrobat is creating more than a couple atomic regions (likely very many). These are several bitmap images, probably along with some other vector type—even if the glyphs stay vector (no transparency affecting them) they are no longer type, but vector objects, if any other part of the copy was rasterized. Acrobat may even be doing you the favor of keeping the text it converted in an invisible layer above the image. (It does this with OCR conversions.)
Text is text, with font attributes applied, and takes up very little room in a document. Text partially converted to bitmaps and vectors can no longer be read in the file as text (exporting, assistive devices, etc.). If, as I suspect, Acrobat is rastering some of the type, it's preserving the text in the file so it can still be read and exported. (This is just a hypothesis on my part.)
All of the various little images in the atomic regions, partially rastered type, plus any overhead garbage Acrobat saw fit to add to your file can increase the file size drastically.
My advice? Either create a PDF and leave the Acrobat watermark alone (there are reasons), or if you have to flatten it—because you're submitting a PDF/X-1a-compliant file for press that needs to have a transparent watermark—flatten the image in Photoshop, then export your PDF.