Computing distances between two empty PersistenceDiagrams

[mbudisic]

A distance (either Wasserstein or Bottleneck) between two empty PersistenceDiagrams should return zero. (Empty diagrams contain trivially the "diagonal" in the birth-death graph, and distance between the diagram and itself should be zero.)

Before this change, the code produces an error when applying either hungarian or _hopcroft_karp algorithms in matching.jl, and the two errors are different.

The pull request performs a simple test at the beginning of Bottleneck() and Wasserstein() functions and returns the distance zero or a zero Matching.

The pull additionally adds a unit test group for this issue.

[coveralls]

Pull Request Test Coverage Report for Build 3552291527

• 6 of 8 (75.0%) changed or added relevant lines in 1 file are covered.
• 16 unchanged lines in 1 file lost coverage.
• Overall coverage decreased (-1.9%) to 94.954%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
src/matching.jl	6	8	75.0%

Files with Coverage Reduction	New Missed Lines	%
src/matching.jl	16	91.88%

Totals
Change from base Build 2382344384:	-1.9%
Covered Lines:	828
Relevant Lines:	872

---

💛 - Coveralls

[mtsch]

Thanks so much, I can't believe I missed such an obvious edge case!

Looks good, but I'd rather it returns a 0.0, so the distances are always Float64s.

[mbudisic]

Ah yes - I didn't pay attention to this. I accepted the suggestions.

Thanks for developing the package, my collaborators and I are enjoying using it!

[mtsch]

Thanks! Looks good now. I'll bump the version and merge/register it now.

[mtsch]

The fix is registered and can now be installed with the package manager. Thanks again!

[mbudisic]

My pleasure!