In general, unless someone's going on a campaign of making small edits, particularly ones that don't fix all of a post, I recommend accepting the edits - if they do in fact improve the post.
Here's my thinking -
A 2k user could make any of these edits and they would bump to the front page - and we want them to because these edits should be reviewed, however much we trust these users.
So, preventing low-rep users from having that impact means that you're blocking them from something that there's no prevention of for others. They're also much more severely limited in the amount of "harm" they can do to the front page because they can only have five suggested edits in queue at a time (or 20 on beta sites). Someone with edit privileges can do as many as they like. They also have a minimum character count to submit an edit suggestion; which high-rep users don't.
The other argument I hear is that they're making more work for high-rep users because they're suggesting edits that reviewers have to approve. That's true but if they really are making the content on the site better, then that should be OK. In many cases, these edits were suggested because the person found the post naturally and had some issues with the existing phrasing, so they submit a suggested edit to make it easier for the next person who comes along.
This is where the judgement call comes in. If it's just one or two suggested edits at any given time, then I don't see it as an issue. Someone who's filling the queue, particularly on beta sites, may want to slow down until they get the edit privileges but, even then, if they are truly improving the content, they're helping the site and those edits will filter down eventually.
By rejecting the edits, you're putting up a barrier that the system doesn't - we block suggested edits on meta sites, so we probably could do it based on post age on main sites... but we don't. Yes, they're adding to the queue but it takes the same amount of work to reject it as to accept it, so once it's in the queue, you're not saving anyone work, you're just telling someone their efforts are unwanted.
We are a community-maintained and edited platform and those efforts should be supported, not seen as troublesome or bothersome by default - they should be judged by their quality, as much as anything else on our sites should be.
So, in short:
Should these edits be rejected? Accepted?
If they only fix some of the issues in a post, either improve the edit or reject and edit depending on the amount of missed fixes.
- If they fixed most of the problems but missed one or two - particularly title edits or tags, improve the edit.
- If they fixed one word but there were several other typos or errors, reject and edit to fix as much as you can see that needs it.
If they fix all of the issues, just accept it and don't pay too much attention to the age of the post.
The caveat here is, if you see someone constantly suggesting 5+ edits every day, that might be a good time to check on the quality of the edits and, if they're sub-par, give some guidance for how they can improve. If you're a mod, you can create a chat room with them and even edit suggestion suspend them temporarily until you've had a chance to get them to see how to improve.
How should the threshold be determined?
Take everything on a case-by-case basis. If you're not recognizing the username in the queue, you can probably just judge based on the quality of the edit and ignore anything else.
I'm glad you point out the rep benefit to new users. Particularly on beta sites 1000 rep gets you a lot and that may be the only way some users have to participate in the site if they're new to the topic but are spending time learning and researching. Letting them help in curation and get a little reward for it is great - just help them do it well if they're failing.