based on my experience (i lost hundreds of hours on a relatively large mod) its a random gamble.
the corrupted areas were all in CELL, WRLD, and possibly NAVI.
in contrast i have an esp that was extensively cut up with tessnip since pre-Ck, re-built post-ck by cut/pasting records from the original pre-ck esp and massively edited hundreds of times using tessnip, and that plugin still works perfectly. that one has many types of records, but nothing world-based.
when i re-built my corrupted mod (replicated most of the base structures for the cells and worldspaces) i could not replicate the tessnip damage (when i deliberately tried some tests using tessnip on a backup copy) even though i did everything the same.
i think the world-based records (references/navmesh etc) are the most prone to the possibility of damage, although there are many users who have used snip even more extensively than i have and they still have no problems with it, so like i said it's a gamble (seems like the odds of getting a corrupted record are low, but a risk nonetheless).