Downsampling images and embedding fonts does not give unexpected results and is a no-brainer. My question is in regards to a couple of other options that I am not sure, without a lot of testing, that I am comfortable with.
These options are:
1. Remove Unneeded Transparencies
2. Flatten Transparent Objects
3. Flatten Form Fields - Not to worried about this one.
4. Discard Hidden Layers
I am nervous about unwanted results in selecting these options as a standard procedure. Are these safe to use on all files, or should they be selectively used upon running into an issue?
It's not easy to make blanket statements when you have seen the many "interesting" files that come out of the woodwork...
In my estimation #1 and #4 are probably the safest to apply to every job. By definition these involve objects that do not alter the appearnace of the document. That said, I would probably give the jobs a quick once over for a window of time.
You may also think about using preflight to determine whether those fixups are needed at all.
I would review everything that was modified by #3 (you can route by fixup applied after preflight).
The appearance of form fields in Acrobat does not always match how the PDF is defined under the hood. This can cause form field flattening to alter the appearance of the document (e.g. a value seen in Acrobat is not seen in the rendered document). I think this is more of an issue for PDFs whose form fields are programmatically added vs those done by an user. However, it's well worth reviewing these sorts of files.
As for flattening... if you have an APPE RIP then you really should not need to do this in most cases (although there are cases in which it will help with performance). If you are sending to PS RIP then it makes sense to flatten.
I would recommend caution and reviewing the flattened documents (as above you can preflight for this).
I know it's not a simple answer but it is what I would do if it was me.