Yes. And No.
That is all.
Strictly speaking, the narrow answer to the question is Yes, the rule against invalidating answers can, but does not always, create questions which can not be opened by any means.
This is one reason why we respondents have a grave responsibility to thoroughly read and understand a question before answering it. We should always discuss clarifications and air issues with the querent before answering.
We often fail our querents in this regard, and I am every bit as guilty as the next person of answering a question that oughtn't be answered.
The failure of my team leads to exactly the problems you mention: questions that can not be improved (edited) without invalidating the answers. This was the topic of my recent Meta question about reminding eager respondents such as myself not to answer questions that really pretty obviously have some kind of problem that needs to be addressed.
Resolution: While not every question that has been closed after an answer is given results in a question that can not be edited; the optimally less than perfect way to solve the problem is simple: learn from the disaster; write a new question that ties the loose ends; move on with the story.
Other Issues & Dilemmas in Order of Appearance:
Not all answers yield paradoxes. I'd argue your analysis is too polar. Just because someone writes a really bad question (poorly conceived, off topic, whatever) doesn't mean the answers will of necessity be bad as well! This is a false conclusion as these answers can still be useful even if the question is not. You can see this all over the forum with questions that are closed or downvoted to oblivion yet have answers that are highly upvoted and often having earned the Green Checkmark seal of approval.
The reasoning here is simple: answers are not the same thing as questions and, as regards their merits, are not as dependent on the question as you assume. You as querent write your question and it is judged on its own merits as to whether it is on or off topic, well or poorly conceived, how well it fits with SE rules and forum customs. I as respondent write my answer and it is judged on its own merits as to whether it a) answers the specific question and / or b) is deemed to be generally good (useful, entertaining, comprehensive, instructional, educational, etc) by others.
Issue 1: A question that is invalid has an OP intent that is invalid (is it possible to assume an OP wrote what they do not intend?) --- This is possible of course! This is why there is a comments section, so these issues can be brought up and resolved. Preferably before answers start pouring in.
Issue 2: Changing an off topic closed question into a valid question will in all cases change the OP intent (you have changed the question topic) --- Not true. Questions can be "off topic" for a variety of reasons. If it's off topic for lacking focus, then focusing the question does not keep it off topic and will not likely change the OP's intent.
Issue 3: Editing an off topic closed question while maintaining the OP intent, can NOT change the question topic; therefore it can never be opened. --- Convoluted and half true. Again, if a question is off topic for lacking focus, edits will (or should) not change the topic. Thus the question can be edited and reopened. If the topic is changed, the question should not be reopened, the edit should either be reverted or modified and the OP should be encouraged to write a new question.
Issue 4: Only by changing the topic of an off topic question can you open it, but to change the topic of any question invalidates answers to that question. --- Untrue. Again, there are many reasons for a question to be closed as off topic. Questions can and regularly are edited and reopened without change of topic. Answers may or may not be invalidated by a change of topic, but best practice assumes they will be and thus the discouragement of rewriting a question via the editing process. This is why, once a poorly written or off topic question gets answers, we encourage the OP to just write a new question.
Issue 5: If this is the case, this seems like a simple criteria to find candidates for deletion. There is no logical reason to consider the validity of answers to off-topic questions. By definition, all answers to off topic questions are “not an answer.” --- Generally, we only close questions that are deemed off topic. Deletion I think is reserved for spam and so forth. Hence "Vote to Close" rather than "Vote to Delete".
The second part is simply untrue: An answer's validity stems from two sources. First, of course, is the question. The other is the answer itself. Answers can be both "not an answer" and also "an answer" simultaneously. In practice, "not an answer" is generally reserved for posts that are really more suited as comments, or that ask other questions or that simply have nothing to do with anything.
Conclusion: while the answer to this question is a qualified yes; the logical conclusion to be drawn from the situation is never to delete answers or questions. In WB, we judge questions by one set of criteria and we judge answers by a completely different set of criteria. This means that a question can be closed, but answers can not. Answers can be broadly useful, and even answer a bad question that is now closed, and yet itself not be an "off topic" answer.
The best options are:
- Avoid this situation in the first place by writing a solid query (onus is on the querent) and by reading the question thoroughly for understanding and addressing any issues (onus is on the respondent);
- Rely on your intuition and VTC a question that looks like it could be a) a good question a/o b) looks like it could be problematic; explain your reasons for closing and offer the OP some assistance, especially newer ones;
- Bring a shovel so you can cover up the mess -- on occasion, a querent might just have to give up on a question the rest of us loused up and start over again.