[feat,perf] Add generalized AMOVA, improve perf

[perf]

Speedup amova analyses by separating the analyses into two groups: shared and
private amova calculations. Shared calculations does not depend on distance
matrix and therefore given the same metadata the same calculations are being
done. New approach avoids this recomputation and speeds up amova replicates.

Improve metadata file reading; handle the cases with non-unique group names,
change metadata representation in memory to reduce the overhead and memory
footprint.

[feat]

Add L-level generalized AMOVA. Change standard AMOVA calculations to
generalized form using inverse variance coefficient matrix explicitly.

[chore]

Update test results to match the new order of printing in amova results csv
file.

Makefile

@@ -203,8 +203,7 @@ DEP := $(OBJ:.o=.d)
 FLAGS := $(CPPFLAGS) $(CXXFLAGS)
 
 
-# Versioning
-VERSION = v0.1
+VERSION = v0.2
 
 ifneq ($(wildcard .git),)
 VERSION := $(VERSION)-$(shell git describe --always)
@@ -267,16 +266,14 @@ clean:
 ## [test]
 # 
 
-.PHONY: test testvg testall
+.PHONY: test test-%
 
 test:
 	bash test/runTests.sh
 
-testvg:
-	bash test/runTests.sh vg
+test-%:
+	bash test/runTests.sh $*
 
-testall:
-	bash test/runTests.sh all
 
 ####################################################################################################
 ## [.activate_module]