What protects the exit is having two of them that both work at the purchase price, and testing that before you bid rather than after.
The rental fallback isn't automatically available. It's available if the property, at your all-in cost, produces rent that covers debt service plus taxes, insurance, maintenance and vacancy at the interest rate you'd actually get on a hold loan. On a house bought at $170k with $50k of rehab, all-in $225k, the rental fallback is real if the market rent supports it and imaginary if it doesn't. Plenty of flips pencil beautifully as flips and cash flow negative $400 a month as rentals, and the operator only discovers that at month nine when the flip won't sell. Run the rent number at the same time you run the ARV, every deal, and write down which ones are single-exit deals. Then treat single-exit deals as requiring a materially wider discount, because you're buying with no plan B.
On ARV going stale: the fix is underwriting to closed comps from a tight window and then discounting for the fact that your sale is months out. Some operators haircut ARV by a few percent per month of expected timeline in a softening market. That's a judgment call, and if you don't make it explicitly you've made it implicitly at zero.
Rehab scope matters more than most people credit, because scope drives timeline and timeline drives both carry and exposure to price movement. A $30k cosmetic job that takes eight weeks has far less market risk than a $90k gut that takes seven months, even if the $90k job shows a bigger spread on paper.
Where I'd push back on your framing: rising insurance costs are part of what's pushing this pipeline wider, and they hit the rental exit directly. A rental fallback underwritten on last year's premium in a coastal or wildfire market may not be a fallback at all. Get an actual bindable quote on the specific address, not a rule of thumb per thousand of value, before you count that exit as available.