There's no reason an SCS time distribution couldn't be used with any reasonable rainfall amount. After all, a 200 year storm in one area may be less than a 100 year storm in another.
The standard SCS/NRCS rainfall distributions are commonly used for everything from 1-year to 500-year events. However, with wider availability of local Precipitation Frequency Data from NOAA, NRCC, and other sources, there is growing interest in generating local synthetic rainfall distributions based on the local PFD data. Although the local curves are often close to the traditional SCS/NRCS curves, they sometimes show significant differences in the peak intensity, and even a variation of peak intensity with return period.
In the end, it's a matter of what your local stormwater regulations allow/require. Although most agencies continue to use the standard SCS/NRCS curves, I expect to see more use of site-specific synthetic rainfall distributions as time goes on.
This trend is fueled by the available of updated PDF data from NOAA, and the development of regional PDF sources such as the Northeast Regional Climate Center.
why 200 year? I have only seen storms larger than 200-year used for analysis of Corps levees and dams. Levee design (except for interior drainage) is generally based on river flows, not on a storm runoff analysis. For the dams, typically the PMF which probably does not use the SCS Type I storm.
By the way, it is more likely that a type II or type III storm would be used for a 200-year analysis.