Bug: segfault with set_abundances

[luizirber]

I was working on a script and triggered a segfault by setting .track_abundance = True and then calling .set_abundances. For now a test reproducing the bug, working on fixing it.

Checklist

• Is it mergeable?
• make test Did it pass the tests?
• make coverage Is the new code covered?
• Did it change the command-line interface? Only additions are allowed
  without a major version increment. Changing file formats also requires a
  major version number increment.
• Was a spellchecker run on the source code and documentation after
  changes were made?

[codecov]

Codecov Report

Merging #745 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #745   +/-   ##
=======================================
  Coverage   88.59%   88.59%           
=======================================
  Files          30       30           
  Lines        4727     4727           
  Branches       46       46           
=======================================
  Hits         4188     4188           
  Misses        536      536           
  Partials        3        3

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6ca4406...5f8d0ab. Read the comment docs.

[luizirber]

This was a bit more nuanced than expected. But it boils down to these assumptions:

1. This is something that a user would do:

   sig = sourmash.load_one_signature(utils.get_test_data("genome-s12.fa.gz.sig"),
                                     ksize=30,
                                     select_moltype='dna')
   mins = sig.minhash.get_mins()
   mins = {k: 1 for k in mins}
   
   new_mh = sig.minhash.copy_and_clear()
   new_mh.track_abundance = True
   new_mh.set_abundances(mins)

   I will say yes, because seemed intuitive when I did it (and found the bug).

2. If you have a minhash WITH abundances, and then set track_abundance = False, it should still work (but dropping the abundances) instead of throwing an error.

3. If you have a minhash WITHOUT abundance, and then set track_abundance = True, it should throw an error (since we have no idea what the abundance of the hashes already there were).
   Caveat: if the minhash is empty, then allow it. This is especially useful with

   mh = MinHash(1, 4, track_abundance=False)
   new_mh = mh.copy_and_clear()
   new_mh.track_abundance = True

   (and was the use case that triggered this bug)......

And thanks @taylorreiter for helping figure out the behaviors for each assumption =]

[luizirber]

Ready for review and merge @ctb @olgabot

[ctb]

LGTM!