Low power rockets, made from thin-wall paper tubing and other lightweight materials and usually using plastic parachute canopies and shroud lines attached with small stickers or tape, really can't handle high deployment velocities and strong opening shocks. Mid power rockets with sewn nylon parachutes, longer and usually stronger recovery harnesses and somewhat stronger structural materials may be able to withstand slightly higher velocities and stronger shocks, but I would still stick to the 20-25 fps limit for them whenever possible. To fly well on commonly-available E, F and G motors, MPRs still need to be built fairly light, and that factor lowers their DV threshold.
High power rockets are designed for recovery deployments under a broader range of conditions and deployment techniques and often use stout harnesses/shock cords and parachutes that are constructed very robustly. As I mentioned before, you usually have more latitude with them.
As both Handeman and I mentioned, there are many variables to consider here. But if you want a simple rule of thumb, a ballpark figure that is generally true of most designs in most situations, stay below 25 fps and definitely do not exceed 30 fps for low power and most mid power rockets. As I said, high power rockets have a wider variety of component materials, construction techniques and deployment methods, and consequently that category of rockets has a broader range of acceptable DVs. I have no doubt that a well-built HPR probably can handle a 50 fps deployment velocity, but that speed will easily strip the chute from a low power rocket and will likely produce a zippered airframe in a mid power rocket built with paper tubing, or a cut shock cord in a fiberglass one.
Let me repeat that these are broad guidelines, though. With LPRs and MPRs, it is important to nail the motor delay to within a few percentage points of ideal in order to insure a low-speed deployment and a relatively gentle parachute opening. With rockets that are designed to fly with motors in the A-F and most of the G motor range, you may not be able to use certain motors because their available range of delays and delay adjustments simply will not meet the needs of that particular rocket. For example, I really cannot launch an FSI Orbit on an AeroTech E11 reload, because the only available delay for it is 3 seconds, which will always be too brief for a safe recovery system deployment in this particular rocket. You can use a delay adjustment tool to shorten a delay, but you cannot use it to lengthen one. A flyer can get around this problem by replacing motor-initiated deployment with electronic deployment, but this may not be an option for some designs, especially rockets that are launched on A motors and some B motors.
Let me repeat one more time that these are general guidelines, valid for most rockets in most launch situations. There can be any number of special cases that are able to exceed these ranges, though.