I know that preprints deposited into a preprint server should be in a polished state. However, what should happen once it starts the review process? Should those copy-edits be updated on the preprint? What about updates when replying to reviewer's comments? What if the manuscript is rejected and I am rewriting the manuscript to be resubmitted to another journal in another format?
In principle, anything that happens to the content of a paper before printing (ie, formal publication) should be reflected in the pre-print. So yes, if you correct a bug while the paper is in review, or revise the paper for a new submission, then you should update the preprint.
However, some journals object to authors' posting post-copy-edited (or even post-refereed) revisions. And posting a revision while a paper is under review could interfere with the reviewing process. When in doubt, ask your editor.
Frequent updates may earn you a reputation for being sloppy, especially on a system like the ArXiv that publishes the preprint's revision history. (Why didn't you fix those bugs before you uploaded the first time?) But that's still better than leaving a buggy preprint out in the wild, thereby earning you a reputation for not even knowing (or caring) that you're sloppy. The right answer, of course, is to debug your papers before you post them!