MAINT: hardcoded skbio metrics to temporarily work around NaN to 0 implementation

[hagenjp]

python 3.9 updates.

1. removed all references to 'kulsinski' as it has been removed from the valid beta diversity metrics in skbio (skbio issue #2009)
2. switch out kulsinski for euclidean for the usage example

[hagenjp]

1. added hardcoded metrics that were previously accessed from skbio
2. implemented corresponding tests for these metrics sourced from skbio

[lizgehret]

tested locally, these changes look reasonable. thanks @hagenjp and @cherman2!