We are finding that with floating point tables we would like to use a mixture of categorical and continuous colortables. The current 5.0 code picks one or the other based on raster map type, although it seems that the categorical table processing is quite happy with floats as well as integers.
Perhaps it would be better to base the choice on two tests.
1) If the number of categories is > 0 (for int or float raster) then use the categorical style by default.
2) Add an over-ride parameter to the colortable block that would force a continuous style colortable even if categories are defined.
Cheers,
John Harrop