First, I wouldn't advise canceling a score that strong. You can get in anywhere with a 730, and even if you might do better later, I find it hard to imagine an advantage to canceling it. It can actually help you to perform better on a retake if you know you already have a very strong score 'in the bank' -- then you might feel less pressure on test day. And if you keep that score, then you have something you could use in an application no matter what happens later on.
As for diagnostic tests: if you'll be taking a practice test, you should know why you're doing it. There are really only two main reasons to take diagnostic tests:
- to assess yourself. Note that you don't need to assess yourself, because you have six consistent official test scores. You're almost certainly at a Q49-Q50 and V42-V44 right now, and a 750-760 overall. If you could take a test tomorrow, that's probably the score you'd get. You'd only need to assess yourself if you decide to experiment with new methods or techniques, or if you invest some time improving your conceptual understanding. Then you might want to see how much of a difference that has made, but there'd be no reason to take another test for assessment purposes until you've done that, so not for at least three weeks;
- to practice test-taking strategy, mainly pacing strategy. This might be important for you, from your description of test day -- had you been able to complete the test at your normal ability, you almost certainly would have seen a Q49 or Q50 (and probably a V44, though it's easier to improve Q pacing than V pacing).
Often when high-level Quant test takers have difficulties with pacing, it's because they aren't willing to move on from questions when their solutions aren't going anywhere. At your level, every question will appear at first glance to be solvable. But in math, it's usually true that if you don't see how to solve something within a minute or so, it could take all day -- you're missing something, and there's no way to tell how quickly you'll figure out what you're missing. If that sounds like the issue, you'll want to practice, in a test-like setting, that kind of decision-making: deciding when to invest more time in a question, and when to save that time for later in the test when you know you'll be able to make profitable use of it. Since you are at a Q49-Q50 level, you're going to see some extremely hard questions on an adaptive test, and getting a Q51-level question wrong really isn't going to hurt you. You're not supposed to get many of those questions right unless you're a Q51-level test taker, so you should be perfectly willing to move on from very hard questions unless you can see how to answer them.
If that is your issue, I suppose you could take company tests
only to become more confident making decisions mid-test, about when to invest more time in a question, and when to move on. You'd just be practicing being disciplined in a test-like setting. But if you're using a prep company test for that purpose, that's the only purpose you'd be using it for. You'd want to ignore any score you get, because company tests don't produce reliable scores.
Pacing issues can also arise simply because a test taker is using slow methods, or is slow at arithmetic or algebra, and is simply taking more time than necessary on a lot of questions. If you think that might be an issue, you might want to learn new ways to think about some of the core GMAT math concepts. I've written a set of books that teach the fastest approaches I know of to high-level GMAT math (most of the methods taught by prep companies are fast, nor are they flexible enough to solve very hard questions), and if you'd be interested in any of that material, get in touch at the email address in my signature, and I'll let you know how you can buy them.
If you do invest time trying to improve your true ability, or if you do practice pacing using company tests and want to work out how much of a difference that has made to your GMAT ability, you might want to use a couple of official tests one or two weeks before test day. In your situation, I'd definitely try to go back to whichever tests you did the longest time ago -- hopefully those were tests 1+2, since those have the biggest question bank (so are least likely to deliver familiar questions). If you've largely forgotten the questions from those tests, they'll be your most realistic tests by far (and so are best to save for when you'd find a realistic assessment most valuable). If you do retake a test, and if you recognize a small number of questions, just be sure to pretend they're new to you -- take as much time on them as you would if you'd never seen them before, and only get them right if you'd get them right if they were unfamiliar.
Lastly, I'd disagree with the post above that says bad (or 'not ideal') questions on prep company tests are a "feature, not a bug". Tests are a poor tool to use to practice content anyway (it's much better to do that practice outside of a test setting, where you can, say, practice a single topic all at once, or choose what question to do next based on any weaknesses you discover). But it's also not a good idea to practice from inferior questions when there are good questions you could use instead. Bad questions are bad for a reason. A lot of 'hard' prep company questions are hard because they test things the GMAT simply doesn't test. Test takers gain nothing from studying those questions. Many prep company questions are hard because they are far more complicated than any actual GMAT problems. Those questions emphasize the wrong way of thinking about GMAT math -- it's not a test of how much algebra you can do. It's a test of how well you think about mathematical concepts, and any questions that do not reward the test taker for conceptual thinking are misleading, and are doing the test taker a disservice.
No matter what you do, you're in a very good position to get a better score on your next test. Good luck!