Provide gamma_clipped for Random Distributions

[anandtrex]

It would be nice if PyNN provided a 'gamma_clipped' distribution similar to the 'normal_clipped' already available for random distributions.

[apdavison]

Could you explain the use case you have in mind here?

(My personal feeling is that is is bad practice to use clipped distributions. The normal_clipped distribution was included mostly for backwards compatibility, since many published models use this, clipping at zero to avoid negative values. In such a case, the plain gamma distribution would probably be better, since it always gives positive numbers.)

[anandtrex]

I was specifically thinking of having a way to provide a maximum value when drawing from a gamma distribution, since for many quantities, extremely large values aren't biologically realistic (and also lead to pathological network behaviour). Setting a minimum value would be nice, but is probably not as essential.