Read 2D chemistry probe species after locating the probe

[sbryngelson]

In the 2D branch of s_write_probe_files the species partial densities
were read from q_cons_vf(j-2, k-2, l) before j, k and l were computed
for the current probe and outside the ownership check, so they used the
previous probe's cell or, for the first probe, uninitialized indices.
A gfortran debug run of a 2D chemistry case with probes aborted on an
out-of-bounds index.

Move the read inside the ownership branch after the indices are set,
as in the 1D and 3D branches.

---

Acknowledgement

• I confirm this PR meets the above expectations and reflects my own understanding and real-world context.