Fix bug in grdblend when setting -I that was in fact ignored and got its value from first grid.

[joa-quim]

Before (result obvious wrong as it didn't respect the -I and the blended -R was wrong too)

grdmath -R0/10/0/10 -I2 X = G1.grd
grdmath -R5/15/0/10 -I1 X Y = G2.grd

grdblend G1.grd G2.grd -GG3.grd -I1
grdblend [WARNING]: File G1.grd has different increments (2/2) than the output grid (1/1) - must resample

grdblend G1.grd G2.grd -GG3.grd -I1
grdblend [WARNING]: File G2.grd has different increments (1/1) than the output grid (2/2) - must resample
grdblend [WARNING]: File G2.grd coordinates are phase-shifted w.r.t. the output grid - must resample
grdsample [WARNING]: Output sampling interval in x exceeds input interval and may lead to aliasing.
grdsample [WARNING]: Output sampling interval in y exceeds input interval and may lead to aliasing.

grdinfo G3.grd
G3.grd: Title:
G3.grd: Command: gmt grdblend G1.grd G2.grd -GG3.grd -I1
G3.grd: Remark:
G3.grd: Gridline node registration used [Cartesian grid]
G3.grd: Grid file format: nf = GMT netCDF format (32-bit float), CF-1.7
G3.grd: x_min: 0 x_max: 30 x_inc: 2 name: x n_columns: 16
G3.grd: y_min: 0 y_max: 20 y_inc: 2 name: y n_rows: 11
G3.grd: v_min: 0 v_max: 0 name: z
G3.grd: scale_factor: 1 add_offset: 0
G3.grd: format: classic
G3.grd: Default CPT:

With this PR

grdblend G1.grd G2.grd -GG3.grd -I1
grdblend [WARNING]: File G1.grd has different increments (2/2) than the output grid (1/1) - must resample

grdinfo G3.grd
G3.grd: Title:
G3.grd: Command: gmt grdblend G1.grd G2.grd -GG3.grd -I1
G3.grd: Remark:
G3.grd: Gridline node registration used [Cartesian grid]
G3.grd: Grid file format: nf = GMT netCDF format (32-bit float), CF-1.7
G3.grd: x_min: 0 x_max: 15 x_inc: 1 name: x n_columns: 16
G3.grd: y_min: 0 y_max: 10 y_inc: 1 name: y n_rows: 11
G3.grd: v_min: 0 v_max: 0 name: z
G3.grd: scale_factor: 1 add_offset: 0
G3.grd: format: classic
G3.grd: Default CPT:

Run the tests and this doesn't seem to cause any failures, but our testing system is very broken, at least for me where I can't use the GMT_DATA_SERVER=static and where the CI give different results in all 3 OS.

So, please test this too.