Smarter pattern matching for Tetra

[zeddie888]

From testing, it looks like Tetra is pretty susceptible to false positives (i.e. false stars).

Currently given a 4-star pattern, Tetra just returns immediately if it matches some catalog pattern. Also LOST's version of Tetra uses edge ratios for pattern construction, which is a simpler scheme but might be more prone to false positives?

Some ideas:

• Look at how the original TETRA did pattern construction, I remember it being a lot more steps at least
• Go through multiple 4-star combinations to find matches, accept the identification with smallest error

[markasoftware]

Another simple improvement is to just not match if there are multiple matches all with low error, because there's no way to tell which one is right

[markasoftware]

Another one is to check that the pattern didn't get flipped, which is called the spectrality check in pyramid