Envelope bounds with Index CRSs

[Piero Campalani]

While a first assumption was that each nD grid cell was filling a non-zero support (footprint), it might be better (and less misleading) to let them have 0D support, so that the gml:Envelope and gml:GridEnvelope coincide.

# 1D support
cell     0     1     2     3
         x-----x-----x-----x
CRS   o-----o-----o-----o-----o---> i
      0     1     2     3     4 
ENVELOPE = [0,4]
GRID_ENVELOPE=[0,3]

# 0D support
cell  0     1     2     3
      x-----x-----x-----x
CRS   o-----o-----o-----o-----o---> i
      0     1     2     3     4 
ENVELOPE = [0,3]
GRID_ENVELOPE=[0,3]

Petascope needs a fix in case the latter option is chosen.

[Piero Campalani]

Fixed in changeset:ceb3a38.
Coverages with an indexed CRS, need indexed coordinates, reflecting the internal grid CRS.
It is misleading (and wrong I'd say) what showed in the first option of the ticket summary.