If I understand correctly, P2SH-P2WPKH outputs are simply anyone can spend outputs in BCH.
What would have happened to BTC stored in such SegWit outputs BEFORE the fork, assuming that those addresses had sent several transactions also BEFORE the fork.
In this case, anyone can scan the blockchain and "steal" the bitcoins, despite there being no fault of the person who had those bitcoins.
According to BIP 141, talking about the Bitcoin chain, it'd still trigger "Witness program" and the funds would be stored safely in the SegWit address -after the fork.
Talking about the Bitcoin Cash chain, the miners can redeem SegWit outputs, (both Bitcoin Cash sent to SegWit addresses and Bitcoins in SegWit addresses pre-hardfork.) However those addresses need to have exposed their Witness program, or the outputs can't be redeemed.
Following the Bitcoin Cash forks, any segwit outputs (either existing before the fork or created after the fork), can be spent by anyone (but usually a miner because such transactions are non-standard).
However, since Segwit had not activated prior to the Bitcoin Cash fork, any Segwit outputs at that time on Bitcoin could also be spent by anyone.
Following Segwit activation, Segwit outputs from before the activation can still be spent from but have to follow the segwit rules.
