[raster] AddRasterConstraint does not check arguments upfront

[strk]

I was confused by AddRasterConstraint failing with a vague message "could not add any constraint". Improving the error messages with r12722 I found out I had a typo in the column name, but still the AddRasterConstraint function kept calling each constraint-specific method rather than immediately giving up.

It would be nicer if the existance of the target table/column and its readability would be checked upfront instead, giving a more human-readable message on failure.

Dunno how safe it would be to backport this (togheter with r12722) back till the 2.0 branch

[robe]

strk actually is this a 2.1 issue also I suspect so so switching this to 2.1.9. We can switch to 2.2.1 (after we relase 2.2.0) if it only affects 2.2+

[robe]

Looks like already fixed in 2.2.7 and above.