fail. This fixes it by sorting out the auto increment identifier

Due to changes in BaseFeatureAdaptor we need to save the state of
meta_coord otherwise a number of tests will begin to fail. Same is
true of xref.

* dev:
  update attrib_type table, using longer varchar for 'code' column

if attribute already exists, not need to scream warnings

* dev:
  no need leaving unliked entries when removing slice, remove any attached assembly entries

when removing slice, remove any attached assembly entries

if this is the last feature of its kind, delete corresponding meta_coord entry
if it was the longest feature of its kind, update meta_coord to second longest feature

Thanks to Nathan and Gareth for pointing this out. Due to the Git migration
we now have to use ensembl-funcgen for the regulation checkout.

* dev:
  added option $filename for get_FTP if filename is specified, will save source to file rather than in an array good for big downloads

if filename is specified, will save source to file rather than in an array
good for big downloads

returns the attribute object associated to a given statistic
updated POD documentation

We can no longer use the REST endpoint http://www.ebi.ac.uk/gxa/dataExport/organisms so
have switched to using their FTP site and some regex patterns against the file
names. Whilst not as nice as the rest endpoint it does seem to work

if ignore release is set for the main xref, it should be inherited by the dependent xrefs as well

unique constraint on _attrib tables, there should only be one value for a given object and attrib_type