dens() function gives NaNs when unimix object includes pointmass

[mengyin]

For example:

g=unimix(c(0.5,0.5),c(0,0),c(0,1)) dens(g, 0.5)

will output NaN.

I think the reason is that in function comp_dens.unimix(), dunif(d, m$a, m$b, log) produces NaNs when m$a==m$b.