Addition of utility classes for use in the Ensembl API. The first are shortcuts for asserting object/ref types and the second is a class which removes boilerplate from working with databases.

on new species (i.e. species without previous stable IDs).

Untested at the moment.

-  @seqeds = sort { $b->start() <=> $a->start() } @seqeds;
+  # HACK:   The translation ENSP00000420939 somehow makes the next line
+  #         bomb out ($a or $b becomes undef) if the start() method
+  #         is used.  I haven't been able to find out why.  It has 10
+  #         Selenocysteine seqedits that looks correct.
+  #         /Andreas (release 59)
+  my @seqeds = sort { $b->{'start'} <=> $a->{'start'} } @seqeds;

Undefined value warnings on $species_suffix usage and lowercasing input to allow for matching on registry stored aliases

mapping and other things specific to the Ensembl Genomes people.

Might have broken Ensembl stable ID mapping (in which case I clean it up
later).

transcript had to be fetched from the database.  If this happens, the
user is likely working in a translation-centric way, i.e. not going
through transcripts.  If the reference is weakened, the reference to
the transcript will be undef after the end of the code block, as the
transcript passes out of scope.

before using it.

set_config now only test 0 wsize/skip_zero conflict for wsize passed as params
added some more pod

=head2 load

  Arg [1]       : Boolean $load_xrefs
                  Load (or don't load) xrefs.  Default is to load xrefs.
  Example       : $translation->load();
  Description   : The Ensembl API makes extensive use of
                  lazy-loading.  Under some circumstances (e.g.,
                  when copying genes between databases), all data of
                  an object needs to be fully loaded.  This method
                  loads the parts of the object that are usually
                  lazy-loaded.
  Returns       : Nothing.

=cut

Simplify cdna_start() and cdna_end() by use of transcript().

Transcript.pm.  Avoids littering the code with calls to weaken().

Scalar::Util::weaken() to avoid memory leak.

* Add a transcript() method to the Translation object to retrieve its
  Transcript.

* Modify seq() in Translation so that it temporarily swaps out the
  canonical translation of the transcript if needed.

replace the Bio::EnsEMBL::Collection modules in release 60.

  The is_current attribute was set to 1 when the transcript had it set
  to 0.  There's a difference between undef and 0...