SDL Trados Studio 2017.

I have a project with a Project TM that uses paragraph segmentation, which I'm beginning to regret (it's good in that I can always split segments into the right units, but I have to do a lot of it to make the project TM useful; if I'd gone with the default segmentation rules, there are always useful segments that are so split that I CAN'T join them. But it is a lot less work...)

Anyway, some of the project is translation, some is editing. I want the edits to reach the TM, so I've started aligning them. To make that useful, I did a lot of segment splitting.

Which creates a new problem: to use those edits, I open the original source document for translation. But here the problems arise:

1.) I have to do the same splits; which takes a long time, but otherwise the segments don't populate.

2.) I get suggestions in the Translation Results window, but they don't get entered, and they don't come up in the same order as they do in the paragraph, and the paragraphs are long enough that it's a pain to find them all to enter them.

2.) When the splits are done, I have a document identical to a previously translated document, from the TM's point of view. It ought to just run down through TUs, populating the target segments. But it doesn't. First of all, it takes a LONG time to search for each TM match. Which is ridiculous, because I have a power user computer and a pretty darn small project TM...

In short, can anyone suggest a better way of doing this? And a reason why the TM search is so slow?

Thanks in advance