Note: based off of #2469

- Rewrites `build_tree`, `find_lca` functions as `LineageTree` class.
Using same tests, we produce same results
- Adds `ordered_paths` method to produce ~ordered lineages from tree for
LINgroup ordered output.
- Removes `num_bp_assigned` column because it was artificial anyway (our
counts are all assigned at the genome level) and we're not trying to
replicate a format exactly, as we are with `kreport` output, where we
have this column.

Note that LINgroups will not be ordered _absolutely_, as there will be
some stochasticity as we descend the dictionary. Instead, related
subpaths will be grouped.

e.g. two potential outputs: 

```
LINgroup_name	LINgroup_prefix	percent_containment	num_bp_contained	num_bp_assigned
lg3	2;0;0	1.56	192000	0
lg1	0;0;0	5.82	714000	0
lg2	1;0;0	5.05	620000	0
lg3	1;0;1	0.65	80000	0
lg4	1;0;1;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0	0.65	80000	80000
```

```
LINgroup_name	LINgroup_prefix	percent_containment	num_bp_contained	num_bp_assigned
lg2	1;0;0	5.05	620000	0
lg3	1;0;1	0.65	80000	0
lg4	1;0;1;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0;0	0.65	80000	80000
lg1	0;0;0	5.82	714000	0
lg3	2;0;0	1.56	192000	0
```

In these examples, the `1;0`.. paths are always grouped together, but
may come before or after the `0;0` and `2;0` groups