The seasoning inconsistency you ran into is real and underdiscussed, and the detail about three different answers from the same institution is exactly the kind of thing that costs people money.
For anyone reading this who does not know the term: seasoning is the minimum amount of time a lender requires you to have owned the property before they will refinance it. The clock start matters as much as the length. Deed date, funding date, and certificate of occupancy can be weeks apart, and in a renovation deal those weeks carry holding costs, which are the ongoing expenses of owning a property before it generates income, things like loan interest, insurance, and utilities.
Your $2,100 gap came from a two-month mismatch between the seasoning period you planned for and the one the lender actually applied. That is a modeling problem, and the fix is to build the question into your lender due-diligence checklist before you close the acquisition, not after. Ask for the seasoning policy in writing, ask which event starts the clock, and confirm whether that answer applies to the specific product you are using, because DSCR loans (loans sized on the property's rental income rather than your personal income) sometimes have different rules than conventional products at the same institution.
One thing worth flagging from the strategy guide: BRRRR's refinance step is the most rate-sensitive part of the cycle right now. You got a clean near-full recycle at your numbers, which is harder to achieve in the current environment. That outcome depended on your after-repair value landing where you modeled it. Conservative ARV estimates are carrying more weight than they used to.
What lender type are you planning to use on the next one, a DSCR product again, or are you shopping something different for the refi leg?