LIA correction may introduce no-data

sin(LIA)=0 is not a nodata

Subtask: #165 (closed)

When sin(LIA) map is applied to beta° images, it may introduce 0's that are interpreted as no-data.

We need to distinguish these cases. As 0 is already a no-data value at the end of orthorectification, it seems better to continue with it.

Correction considered

if beta == 0:
    return 0
elif sin(LIA) == 0:
    return lower_signal_value
else:
    return max(lower_signal_value, sin(LIA) * beta)

IOW, the final BandMath shall become: exp = "im{beta}b1 == 0 ? 0 : max(lower_signal_value, im{sinLIA}b1 * im{beta}b1)"

LIA Support applications & S1Tiling shall correctly propagate nodata