Remove deprecated doc and usage language from grdview

[PaulWessel]

There is no need to explain about the deprecated way of specifying 3 grids to drap (red, green blue) since we stopped doing that for grdimage a long time ago. grdview will still check if given 3 but we don't have to talk about it. The usage and docs now deal with -Gdrapegrid|drapeimage only. We have a script that exercises the deprecated forms -Gr.nc,g.nc,b.nc and -Gr.nc -Gg.nc -Gb.nc and that works fine under the hood without any advertisement.

I left the special option -z for passing a single image from GMT.jl. Suggest @joa-quim play with removing that. Since the KEY need G or I perhaps some work needs to be done in GMT_Encode_Options?

Improved the doc layout for -G and -Q as well as their synopsis.

Here is -G:

Here is -Q:

All tests pass so I think this is mostly a documentation PR.

[joa-quim]

Not sure you can drop the case z: like that. grdview has a KEY zI(, but no equivalent GI(,

[PaulWessel]

Yes, saw that so leave for now. Solution is to learn in GMT_Encode_Option that the virtual single file is an image, and then change type from ? to I or G.
But grdimage also accept image or grid and it just does <G{ and never sees an I even though we detect that in the code, no?

[joa-quim]

Yep, that magic is apparently working. I have this Julia test

I = mat2img(rand(UInt8,89,120,3), proj4="+proj=longlat +datum=WGS84 +no_defs");
grdview(rand(90,120), G=I, J=:X12, JZ=5, Q=:i, view="145/30", show=1)

and a bunch of colored random noise draped on a random grid does show up and looks good. So that -z is probably not needed after all.

[PaulWessel]

Try to remove the z KEY and the case 'z': in this PR and see if all is well.

[joa-quim]

Removing the z and its KEY as well as replacing zI( by GI(, all made no difference. The colored trash test still works fine.

[PaulWessel]

OK, did you commit this to the PR so we can approve merge etc?

[joa-quim]

Did commit now. Also replaced the KEY though it's probably not being used.