Minor aesthetic inconsistent behavior for legend entry referring to multiple curves

I really like the fact that I can have multiple curves refer to the same legend entry:

a = hv.Curve(np.arange(10), label='this').opts(color='black')
b = hv.Curve(10-np.arange(10), label='this').opts(color='black')

This means that clicking that entry toggles transparency on all of the referring curves, and enables me to present time series with gaps very easily:

curves = []
for j in range(10):
    X = np.random.rand(10)
    t = np.arange(len(X))+j*len(X)*1.2
    curves += [hv.Curve((t,X), label='a').opts(color='blue', alpha = .3)]

The issue arises when using transparency in individual lines (like in the above example). It seems that all the legend entries are actually over-plotted on the same position, so that the legend entry does not have the same transparency as the line.

Am I missing something?

I just want to add that I understand I can work around this by explicitly cutting the curves using nans, but it does not show the legend when there is a single curve. For more than one, it works nicely. It would just be easier (and simpler?) to not need to do that

# an example a bit closer to my use case 
# where I show some estimates and their confidence intervals

curves = []
i = 0
for lbl in ['blue',"red"]:
    data = []
    t = []
    for j in range(10):
        data += [4*i+np.random.rand(10), [np.nan]]
        t += [np.arange(10)+j*10*1.2, [np.nan]]
    curves += [hv.Curve((np.hstack(t), np.hstack(data)), label=lbl).opts(color=lbl)]
    curves += [hv.Area((np.hstack(t), np.hstack(data),1+np.hstack(data)), label=lbl, vdims=['0','1']).opts(color=lbl, alpha = .3)]
    i += 1