Fix iridas cube parser to accept and convert double precision.

[Shootfast]

The old Iridas cube parser would silently convert double precision values to floats and continue working, wheras the current code will raise an "Invalid color triplets" error.

This patch restores the older behaviour.

[linux-foundation-easycla[bot]]

• :white_check_mark:login: Shootfast / (0ff59512a0f57b1a5486992ca15f0453c7178135)

The committers listed above are authorized under a signed CLA.

[Shootfast]

Actually I've just noticed this problem extends to other lut file formats as well - at least spi1d/spi3d