Subplot with coast -E fails to set correct projection/region in subcalls

[PaulWessel]

While I thought we fixed this a short while ago, apparently not. This example shows the problem:

gmt begin Subplots png
  gmt subplot begin 1x1 -Ff20c
    gmt coast -R-80/-64/-6/14 -B -BWNse+t"Colombia" -ECO+gred -Wthin -Gdarkolivegreen -Slightblue -N1/thick,darkbrown -c
  gmt subplot end
gmt end show

with result

The problem is that coast will get the polygon to be painted red and call plot and clearly the argument passed (-R -J) are not correct.

[PaulWessel]

It is probably similar to #6751.

[PaulWessel]

@joa-quim I think I will have to implement #6751 differently since we may have insets inside subplots as well and hence there can be two separate scales to apply. So I think we need to manipulate two scales: inset and subplot, and then when a module does its map_setup we must scale the map scale by inset * subplot [both of which default to 1]. I will try to get to this on the weekend but by now I have added at least 2 weeks of work to that one weekend so we shall see.