Bug fixes, blend endpoint

• blend endpoint added to the service
• removed lg tables and dictionaries.
• remove p_group table
• store IMGT db version
• Fix XX errors for broad/splits
• Fix loading of IMGT DB version 3130

Shortnulls for exon mode fixed

See PR #183 to fix #161

Ping! Mode and DR Blending

PING mode

• When ping=True alleles in P-groups are included in addition to the alleles in G-groups in the corresponding lg/lgx groups.

DR Blending

• Support DRBX Blending from DRB1

`py-ard` REST Service

py-ard REST Service

• REST service with /redux, /validate, /mac endpoints
• Run in a Docker container

BugFix for HLA Prefix

See #165

Reduce to Shortnulls

Implement shortnulls behavior with new config variable set to True by default.

This will address things like DRB4*01:03N and DRB5*01:08N both in terms of accepting them as valid and also by expanding them as appropriate to the list of longer alleles that have the same expression character.

DRB5*01:08N is acceptable by WHO rules since all suballeles have N.
DRB4*01:03N is acceptable by WMDA (but not WHO) rules but this needs to be handled for things like HF analysis by expansion to something valid (vs rejection) to avoid bias

When there are multiple expression characters in the same group defined by a shortname create multiple shortnulls for the same WHO group. e.g.

{'N', 'L'}  A*02:01
{'L', 'Q'}  A*02:01:01
{'L', 'Q'}  A*24:02
{'L', 'Q'}  A*24:02:01
{'L', 'Q'}  A*31:01
{'N', 'Q'}  A*31:01:02
{'N', 'Q'}  B*15:01
{'N', 'Q'}  B*15:01:01
{'N', 'Q'}  B*44:02

See #154 and #155

Support shortnull

Implements shortnulls behavior with new config variable set to True by default.

This will address things like DRB4*01:03N and DRB5*01:08N both in terms of accepting them as valid and also by expanding them as appropriate to the list of longer alleles that have the same expression character.

DRB5*01:08N is acceptable by WHO rules since all suballeles have N.
DRB4*01:03N is acceptable by WMDA (but not WHO) rules but this needs to be handled for things like HF analysis by expansion to something valid (vs rejection) to avoid bias

Summary Table for Batch

Show summary table for failed alleles.

Summary
-------
35 alleles failed to reduce.
| Column  Name    |      Allele      |      Did you mean ?
| --------------- | ---------------- | -------------------------
| r_A_TYPE1       | A*24:09          | A*24:09N
| r_A_TYPE1       | A*33:157         | A*33:157N
| r_A_TYPE1       | A*26:25          | A*26:25N
| r_A_TYPE1       | A*23:19          | A*23:19N
| r_A_TYPE1       | A*24:02:01:02    | A*24:02:01:02L

Fix batch regression error

Fix regression error for MAC reduction #152

Add reduce_2field reduce option

Batch csv pyard-reduce-csv takes in reduce_2field option to reduce 2 fields alleles like B*15:220, DPB1*104:01 and A*02:642