This directory has the source of the autocorrelation used to compute
the contrast sensitivity table.  See 2019A&A...621A..86B for background.
Since we assume EDR2 has a limit of ~20 mag, and radii >5'' are only
relevant when Δm>9, we do two queries, one for objects brighter than 11
with a large radius, one of objects fainter than 11 with a smaller
radius.

To (re-) create mock_pairs.csv, just run

  psql gavo < autocorr.sql

-- this contains psql instrumentation to spit out some pseudo-CSV.

Since a naive pair selection yields a humungous number of results (2.5e9
for the basic cut-at-11 query), we're further filtering in the outer
loop. We're approximating the 99% curve from Table 1 in
2019A&A...621A..86B with

 s = 0.84 (Δm-0.5) + 0.66

-- that's pretty much throughout on the safe side.

From mock_pairs.csv, based on the lookup table in the paper,
make_detectability.py creates data/detectability.csv, a text file of
(source_id, sensitivity) that's then further processed in q.rd

Run make_detectability.py in this directory.