with keeping all AVLs mapped seems the most efficient. Now 1 million
sequences more or less constantly takes 1 minute.

import AVLs from a DMS to another

copied properly (because the sutructure copy via assignation does not
work for structures with a variable size)

correctly enlarged before copying the data) + minor improvements

group to make it writable. Also fixed several bugs when handling AVL
groups.

closed. Obligatory indexers' names now follow the same pattern as other
indexers (columnname_version). Closes #46 and #49.

an AVL is opened (as opposed to created), it is read-only

functions in the obidms files.

separate obiblob.c and obiblob.h files

AVL trees

macros for crc

(note: unretrievable data as implemented)