Exclude MISSING when counting alleles per site

[szhan]

Fix #33

I reworked check_alleles in lshmm/api.py.

I implemented a helper function _get_num_alleles, which has the same logic as check_alleles, in the following two test files:

• test_API.py
• test_API_multiallelic.py

The above sets of tests pass.

I did not make any changes to test_LS_haploid_diploid.py, because there is no testing of FB when mutation rates are scaled by the number of distinct alleles.

Also, I'm only dealing with MISSING here. I'll add NONCOPY to the exclusion set in #31.

[astheeggeggs]

This all looks good to me, defining a function to count the number of non-missing states in the reference, and count that as you track along the genome makes sense.

[szhan]

Is this okay? If so, then perhaps we could merge it before I do some refactoring?

[jeromekelleher]

LGTM

Shall we merge @astheeggeggs?

[astheeggeggs]

Sure, will merge