add methods to < and <= instead of to > and >=

[nsajko]

As documented, the intended way to implement > is to add a method to <. Similarly with >=.

This should solve some of the problems mentioned on issue #1799. However, for the long term (whenever a breaking release of GMT.jl happens), I'd suggest just getting rid of the methods, as discussed on issue #1799.

[joa-quim]

Sorry, but this also removes the G > 10 or I > 10 syntax and that is whole point of the situation. I want to be able to do both
G > 10 or G < 100 (G(rid) or I(mage) is the same for this discussion). And as it is G < 100 is considered valid but G > 10, not.

Users could also do !(G <= 10) to get the same result, but users would say (or find out that they had to do it that way?)

[nsajko]

Sorry, but this also removes

Can you be more specific?

[joa-quim]

If we replace

Base.:>(I::GMTimage{T}, val::Real) where T <: Unsigned = mat2img(collect(I.image .> val), I)

by

Base.:<(val::Real, I::GMTimage{T}) where T <: Unsigned = mat2img(collect(I.image .> val), I)

the multidispach I::GMTimage > ... is gone and I > ... will be a No method .... error.

[nsajko]

Why do you think so? As discussed already, > has a generic definition, falling back to <.

[joa-quim]

Can't say I understand how the fallback worked, but you are right. This works to keep the > and >= syntax.
Thanks a lot

(I'm now down to 31 SnoopCompile invalidations, but that doesn't change a bit in 59494)