If you generate regular stuck-at patterns you will detect hold_time type violations. This is because stuck-at will likely launch a transition and if there is a hold time violation the capturing flop will capture the new data and fail.
Regarding path delay there are things you can do like increasing the abort limit that may help. Also, please refer to technote MG520378 in https://support.mentor.com
Please feel free to open a service request if you need more help.
I think the critical thing is in your wording "stuck-at will likely launch". Synopsys seem to have a fault model that specifically targets short paths and generates patterns to create hold violations. Our AE in the US suggested possibly a user-defined fault model......I'm not that convinced of the value of this anyway but we are often asked to do things we don't agree with :-)
Personally, I think the timing-aware ATPG is of more use, so that's what I'm currently working on.
Thanks for the help