allows testing on patches and similar

Check feature out of destination slice. SimpleFeatureAdaptor debugged in the negative strand slice case

Removed methods get_all_(Genes|Transcripts|Exons|Attributes) which weren't doing the correct thing in case of slices and features spanning the origin of replication

using current_schema_no_cvs everywhere now

when running stable id mapping, exons are projected onto reference chromosome
added method in GeneScoreBuilder and EnsemblGeneGeneric to distinguish genes depending on their initial slice
added method in EnsemblExonGeneric to distinguish exons according to the mapped transcript they belong to

The big post-merge alt allele commit. Changes to alt_allele table, new tests, and a couple of new objects.

now closing file with same name as well

- sub_Slice_same (does not weaken adaptor)
- seq_region_Slice_same (does not weaken adaptor)
- get_seq_region_id_same

This means that CoordSystem, DensityType, Map::Marker, Map::Ditag, Map::Qtl,
MiscSet, OntologyTerm and SeqRegionSynonym have now a fast constructor available.

Accordingly, new_fast is removed from Analysis, Feature, DataFile, DBEntry, RepeatConsensus,
Translation and UnmappedObject together with the import of weaken and isweak from
Scalar::Util.

Added error tolerance to cache copy in get_all_Transcripts. Major issue with original method returning undef instead of \[]. Not changed for legacy support.

currently not compatible with most existing tests

method to use is get_seq_region_id()

satisfies perl critic

Replaced all instantiations of pairs of circular slices with a call to the
above method.

An option from the command-line (-current-schema-no-cvs) allows to load the
schema from the file system instead of fetching it from the CVS repository.

we do not want to retrieve all symlinks of the slice, just the slice itself

constructor of its parent (Storable) which is neglecting a lot of the UnmappedObject specific
attributes which instead are assigned and dealt with in the UnmappedObject constructor.
This hasn't been an issue so far as there doesn't seem to be any code calling this new_fast
implementation.

Improvements to documentation, describing the presence of forward and reverse strand features in the results of get_all_* methods.