Correct (or remove) toptaxonXname fields in observation

[regetz]

The observation table has denormalized columns toptaxon1name, toptaxon1name, ..., toptaxon5name, which should contain the names of the top 5 taxa based on cover (presumably using max cover across strata?). However, they don't always seem to be populated as expected. Work here would be either (a) correct these everywhere, or (b) remove them from the DB schema altogether.

Example:

select authorobscode, toptaxon1name, toptaxon2name,
       toptaxon3name, toptaxon4name, toptaxon5name
  from observation where observation_id = 10404;
/*
-[ RECORD 1 ]-+--------------------
authorobscode | INW07566
toptaxon1name | Pinus contorta
toptaxon2name | Spiraea betulifolia
toptaxon3name | Spiraea betulifolia
toptaxon4name | Spiraea betulifolia
toptaxon5name | Lonicera utahensis
*/

[regetz]

Note that we're removing these fields from API responses in #461, but the current issue is about cleaning up the DB itself (either by correcting the values or dropping the columns).