Release 3.2.0

• mmseqs2 replaces blast in sequence search for precomputation, dramatically reducing time needed for PDB precomputation #46
• mmseqs2 replaces blastclust for redundancy elimination clustering. Removes blastclust dependency
• a single run of redundancy elimination with clustering at 90% seq id is now used (instead of incrementally trying different sequence identities)
• extracted utils into new module eppic-util
• pdb to uniprot reference alignment now strictly from SIFTS #188
• all insertion/deletions are now not considered for evolutionary analysis
• no evolutionary analysis performed for peptides (30 default length cutoff). Solves #9, #5
• keeps blastp for user jobs. Mmseqs is not fast for single jobs (unless setup with a server holding the indices in memory)
• user jobs use uniprot japi instead of local uniprot db
• sequence search cache (before blast cache) is now in db (via JPA)
• local uniprot db is now interfaced via JPA instead of raw sql. DB interface now fully JPA
• improvements in precomputation scripts to support mmseqs