Many languages have epenthesis and deletion, but obviously for very different reasons. Is there an OT overview with an account of the most common epenthesis and deletion processes? That way when I work on an OT account of a language with either of those processes, i can refer to comparable accounts given for other languages, to help guide and inform my analysis.
The simple answer is, yes, there are myriads. As far as I know, there is no work that attempts to assemble all or even 50% of the instances of epenthesis and deletion across languages, and discuss what it all means within OT (that would be way too broad). An "overview" would probably mention the best-known trends, and there are a number of such works (although not limited to just insertion and deletion).
A couple of references would be Prince and Smolensky "Optimality theory" and McCarthy "What is Optimality theory". P&S does have the disadvantage that it has no faithfulness constraints, but otherwise this overviews the basic logic of "why do we get epenthesis?" and "why do we get deletion?". McCarthy WIOT gives examples whereby Max and Dep are called on to rein in those two processes (though it is heavy on technology and light on cross-linguistics). McCarthy & Prince "Faithfulness and Reduplicative Identity" lays out Correspondence Theory and exemplifies with numerous insertions and deletions, especially considering "factorial typology".
Such discussions have to factor in the vast wealth of reasons for inserting or deleting – "because this language hates coda consonants", "because they hate inserting vowels", "because that really want to keep the end of the morpheme aligned with the end of the syllable", "because this language wants adjacent segments in the input to be adjacent in the output", "because the language hates short words". There is in fact nothing that is universally irrelevant to the choice, so clearly one cannot say anything useful about why there is insertion and deletion.
Most attention has been given to the question of what gets inserted. Lombardi, for instance, has a paper on the typology of epenthetic vowels, if you want to know why a certain vowel is the epenthetic vowel (for example, why is [ỹ:] never or at least rarely the epenthetic vowel, though you'd have to figure that out on the basis of her approach, she doesn't specifically talk about that possibility). Going back to the very beginning, OT had markedness constraints like *t, *k, *s, and by ordering these constraints correctly e.g. by making *s be the lowest-ranked constraint, you can regulate the nature of the inserted segment. Limiting deletions (for instance, only high vowels undergo syncope, only glides are deleted inter-vocalicly) can be regulated either by the motivating constraint ("high vowels in the context VC__CV are forbidden") or by a faithfulness constraint (vowels in the context VC__CV are forbidden, but deletion of low vowels is even more forbidden). Kager 1999 gives a brief survey of insertion and deletion as driven by syllable structure (typically, no branching or onsets, *Coda, Ons).