The inner loop is always across the model layers while we ensure
that the 'outer' loop always iterates over an index range that is at
least as large as that of the 'middle' loop. This follows from
recent work on compatibility prompted by real field models with the
constant flux analytic aquifer type.
Thanks to [at]tskille for the additional insight.
The protocol for outputting an aquifer's active cell IDs to the
restart file requires that the cells be enumerated according to
their column. In particular, the aquifer's notion of a model's
active cells correspond to a cell ordering of the form
for (i = 0; i < Nx; ++i)
for (j = 0; j < Ny; ++j)
for (k = 0; k < Nz; ++k)
use(X(i, j, k))
instead of the common ordering, sometimes referred to as "natural
ordering"
for (k = 0; k < Nz; ++k)
for (j = 0; j < Ny; ++j)
for (i = 0; i < Nx; ++i)
use(X(i, j, k))
which is employed elsewhere.
This commit adds a special purpose mapping facility which will
translate active cell IDs from natural ordering to the columnar
enumeration. The facility is intentionally bare-bones because we
do not expect to use it outside of the restart writing code, and
only in a limited fashion within that context. We will however
consider adding more features if the need arises.
