Creating a taxon manually can cause invalid snapshot files

[mmore500]

    tax1, tax2 = "hello", "hello2"
    sys = systematics.Systematics(taxon_info_fun, True, True, False, False)
    org = ExampleOrg(tax1)
    org2 = ExampleOrg(tax2)
    org_tax = systematics.Taxon(0, tax1)
    org2_tax = sys.add_org(org2, org_tax)
    org3_tax = sys.add_org(org2)
    org4_tax = sys.add_org(org, org2_tax)
    org5_tax = sys.add_org(org, org4_tax)

produces

id,ancestor_list,origin_time,destruction_time,num_orgs,tot_orgs,num_offspring,total_offspring,depth
3,[1],0,inf,2,2,0,0,2
2,[NONE],0,inf,1,1,0,0,0
1,[0],0,inf,1,1,1,1,1

Because org_tax is not registered through add_org it won't be included in the Snapshot file, leaving dangling ancestor references that segfault when you try to load the Snapshot file back up.

[mmore500]

We might want to hide systematics.Taxon from users? What's the use case for manually constructing a Taxon?

[emilydolson]

Yeah, fundamentally no one should ever construct a Taxon object except through the systematics manager. It was in the tests mostly just to make sure things were working as they should, not because it's a good use pattern.

Is there a way to hide the Taxon constructor from users but not the setters/getters? The reason I left Taxon objects in the documentation is that there's lots of reason to interact with them, just never to directly construct them.