Yes, there are some simple joins already, but I'd have to replace, for example:

delete from testDBRepCats where repCatID = ?
[download]

with something like:

delete r.* from testDBRepCats r, testDBToMCQDB m, testDB t, clients c 
+where r.repCatID = m.repCatID and m.testDigest = t.testDigest and t.c
+lientID = c.clientID and c.drone != '$drone'
[download]

(not checked for validity)
except that's much more complex to chunk (for time-out purposes), because just adding limit 1000 won't maintain data integrity the way the current approach does -- I'm processing 1000 tests at a time, but all the data for those tests is removed before I check time and move to the next chunk, so I'm left with something like:

delete r.* from testDBRepCats r, testDBToMCQDB m, testDB t, clients c 
+where r.repCatID = m.repCatID and m.testDigest = t.testDigest and t.t
+estDigest = ?
[download]

passed to execute_array(), which may or may not be any gain over the current approach, since it's a more complex statement, involving a lot more tables.