Effect of rebalances with disabled MC on LND DB

[GordianLN]

In a convo with @C-Otto we examined how mission control logic, while useful on paper, will prevent you from establishing successful paths based on presumptions that are sometimes at fault: an edge along the way might have pending htlcs that compromise its liquidity but for a second, and be ignored for this reason when trying again just a little later would work, or that same edge might get a rebalance in a few minutes later, to the same effect.
So, assuming an operator didn't really care for the computational load on its node while examining additional paths, what would be the downsides of "routing commando", that is with MC disabled?
I understand this will increase pending htlcs all around, also possibly (marginally?) increasing the risk of force closures... how much impact would this also have on the lnd database? Will it for some reason bloat out of proportion because of testing so many paths?
And if yes, would this be solved/mitigated by deleting ALL payments daily, or by 0.15?

Sneaking in an additional Q: would there be a way to retain MC functionality, but only in a non-penalizing way? For example, rather than ignoring unsuccessful edges, promote instead successful ones, so they are tried first over others, while keeping all unsuccessful ones in the queue, just further on?

[GordianLN]

I thought this would generate more interest...