stable id lookup now also contains an archive_id_lookup table, which stores retired stable ids
also reduced the required memory
for release 74 and 66 species, it ran with less than 600Mb, no need to request 1.5G

switch to using the local as default. otherwise we cannot have a test passing until we have committed to version control

switch to using ENV variables. Otherwise we run the risk of not being able to run the tests in any kind of test harness tool

A number of params are missing their - flag. That means the new XS rearrange does not work (it is a bug not a feature)

- schema version change: 74 --> 75
- addition of source column to transcript table.

Our lookups of primary accessions and display labels with the Xref table was
flawed. Seems MySQL can only use 1 index per query and it is not possible to
create a single index that can handle querying on two columns when either
could be our matching value. Solution is to break the query into two parts
each targetting a single column. MySQL is a lot happier about this and
query times have dropped from 30 seconds on USEast to 18ms (9ms per query).

When a feature is returned by a method in its coordinate system (e.g. fetch_by_dbID)
and it's on a circular chromosome and spans the origin of replication,
Mapper::add_map_coordinates aborts since it's testing the length of the contig
against the difference between chr_end and chr_start.

same as store_mapping_path, but for multiple mappings
a multiple mapping is needed when a single component can be mapped to multiple assembled regions
this uses the '#' operator instead of the '|' for the mapping key

requires a number of updates to methods when adding/fetching mapping_path
methods used to create mapping_keys as name1:version1|name2:version2
if there is no version, do not add ':' either
for add_mapping_path method, do not sort list of coord systems by rank
first argument is the assembled, second argument is the component, regardless of rank

also, when adding_mapping_path, do not expect the method to order coord systems by rank, it will use the first argument as assembled, the second argument as component

works similar to store_assembly, used to check if assembly entry already exists between 2 slices before attempting to store
also corrected typo in store_assembly method, table used is assembly, not analysis

uses code to fetch result

The ID for simple features seem to be poorly defined and repetitive. Better
to supress for the sake of GFF3.

readable format for pipelines, etc.

The idea is good. Consult multiple adaptors as if they are one logical
adaptor with the same query. Thing is that the old version was abused. Let's
stop that now and bring in this nicer new version which does work well