You cannot assign observed to Deterministic nodes. However, what you said about scale of prior (and shape of prior for that matter) is valid. I would suggest you to plug in some number to p and q and see what is the scale of mu using your current prior.
In addition, since the two node is fully observed with a uniform prior:
t = pm.Uniform('t', lower=0, upper=100, observed=sales.t)
cid = pm.Categorical('cid', p=np.repeat(1./sales.shape[0], sales.shape[0]), observed=sales.region )
They are no different than doing just
t = sales.t
cid = sales.region